EDWARD M. PURCELL AND
DAVID J. MORIN
THIRD EDITION
Electricity and Magnetism
For 50 years, Edward M. Purcell's classic textbook has introduced students to the world
of electricity and magnetism. This third edition has been brought up to date and is now
in SI units. It features hundreds of new examples, problems, and figures, and contains
discussions of real-life applications.
The textbook covers all the standard introductory topics, such as electrostatics, mag¬
netism, circuits, electromagnetic waves, and electric and magnetic fields in matter. Tak¬
ing a nontraditional approach, magnetism is derived as a relativistic effect. Mathemat¬
ical concepts are introduced in parallel with the physical topics at hand, making the
motivations clear. Macroscopic phenomena are derived rigorously from the underlying
microscopic physics.
With worked examples, hundreds of illustrations, and nearly 600 end-of-chapter prob¬
lems and exercises, this textbook is ideal for electricity and magnetism courses. Solu¬
tions to the exercises are available for instructors at www.cambridge.org/Purcell-Morin.
edward m. purcell (1912-1997) was the recipient of many awards for his scientific,
educational, and civic work. In 1952 he shared the Nobel Prize for Physics for the dis¬
covery of nuclear magnetic resonance in liquids and solids, an elegant and precise
method of determining the chemical structure of materials that serves as the basis for
numerous applications, including magnetic resonance imaging (MRI). During his career
he served as science adviser to Presidents Dwight D. Eisenhower, John F. Kennedy,
and Lyndon B. Johnson.
david j. morin is a Lecturer and the Associate Director of Undergraduate Studies in the
Department of Physics, Harvard University. He is the author of the textbook Introduction
to Classical Mechanics (Cambridge University Press, 2008).
THIRD EDITION
ELECTRICITY
AND MAGNETISM
EDWARD M. PURCELL
DAVID J. MORIN
Harvard University, Massachusetts
Cambridge
UNIVERSITY PRESS
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town,
Singapore, Sao Paulo, Delhi, Mexico City
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/Purcell-Morin
© D. Purcell, F. Purcell, and D. Morin 2013
This edition is not for sale in India.
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
Previously published by Mc-Graw Hill, Inc., 1985
First edition published by Education Development Center, Inc., 1963, 1964, 1965
First published by Cambridge University Press 2013
Printed in the United States by Sheridan Inc.
A catalog record for this publication is available from the British Library
Library of Congress cataloging-in-publication data
Purcell, Edward M.
Electricity and magnetism / Edward M. Purcell, David J. Morin, Harvard University,
Massachusetts. - Third edition,
pages cm
ISBN 978-1-107-01402-2 (Hardback)
1. Electricity. 2. Magnetism. I. Title.
QC522.P85 2012
537-dc23
2012034622
ISBN 978-1-107-01402-2 Hardback
Additional resources for this publication at www.cambridge.org/Purcell-Morin
Cambridge University Press has no responsibility for the persistence or
accuracy of URLs for external or third-party internet websites referred to
in this publication, and does not guarantee that any content on such
websites is, or will remain, accurate or appropriate.
Preface to the third edition of Volume 2
Preface to the second edition of Volume 2
Preface to the first edition of Volume 2
* CONTENTS
xvii
xxi
CHAPTER 1
ELECTROSTATICS: CHARGES AND FIELDS 1
1.1
Electric charge
1
1.2
Conservation of charge
4
1.3
Quantization of charge
5
1.4
Coulomb’s law
7
1.5
Energy of a system of charges
11
1.6
Electrical energy in a crystal lattice
14
1.7
The electric field
16
1.8
Charge distributions
20
1.9
Flux
22
1.10
Gauss’s law
23
1.11
Field of a spherical charge distribution
26
1.12
Field of a line charge
28
1.13
Field of an infinite flat sheet of charge
29
1.14
The force on a layer of charge
30
1.15
Energy associated with the electric field
33
1.16
Applications
35
VI
CONTENTS
Chapter summary 38
Problems 39
Exercises 47
CHAPTER 2
THE ELECTRIC POTENTIAL 58
2.1 Line integral of the electric field 59
2.2 Potential difference and the potential function 61
2.3 Gradient of a scalar function 63
2.4 Derivation of the field from the potential 65
2.5 Potential of a charge distribution 65
2.6 Uniformly charged disk 68
2.7 Dipoles 73
2.8 Divergence of a vector function 78
2.9 Gauss’s theorem and the differential form of
Gauss’s law 79
2.10 The divergence in Cartesian coordinates 81
2.11 The Laplacian 85
2.12 Laplace’s equation 86
2.13 Distinguishing the physics from the mathematics 88
2.14 The curl of a vector function 90
2.15 Stokes’ theorem 92
2.16 The curl in Cartesian coordinates 93
2.17 The physical meaning of the curl 95
2.18 Applications 100
Chapter summary 103
Problems 105
Exercises 112
CHAPTER 3
ELECTRIC FIELDS AROUND CONDUCTORS 124
3.1 Conductors and insulators 125
3.2 Conductors in the electrostatic field 126
3.3 The general electrostatic problem and the
uniqueness theorem 132
3.4 Image charges 136
3.5 Capacitance and capacitors 141
3.6 Potentials and charges on several conductors 147
3.7 Energy stored in a capacitor 149
3.8 Other views of the boundary-value problem 151
3.9 Applications 153
Chapter summary 155
CONTENTS
VII
Problems 155
Exercises 163
CHAPTER 4
ELECTRIC CURRENTS 177
4.1 Electric current and current density 177
4.2 Steady currents and charge conservation 180
4.3 Electrical conductivity and Ohm’s law 181
4.4 The physics of electrical conduction 189
4.5 Conduction in metals 198
4.6 Semiconductors 200
4.7 Circuits and circuit elements 204
4.8 Energy dissipation in current flow 207
4.9 Electromotive force and the voltaic cell 209
4.10 Networks with voltage sources 212
4.11 Variable currents in capacitors and resistors 215
4.12 Applications 217
Chapter summary 221
Problems 222
Exercises 226
CHAPTER 5
THE FIELDS OF MOVING CHARGES 235
5.1 From Oersted to Einstein 236
5.2 Magnetic forces 237
5.3 Measurement of charge in motion
5.4 Invariance of charge 241
5.5 Electric field measured in different frames
of reference 243
5.6 Field of a point charge moving with constant velocity 247
5.7 Field of a charge that starts or stops 251
5.8 Force on a moving charge 255
5.9 Interaction between a moving charge and other
moving charges 259
Chapter summary 267
Problems 268
Exercises 270
CHAPTER 6
THE MAGNETIC FIELD 277
6.1 Definition of the magnetic field 278
6.2 Some properties of the magnetic field 286
CONTENTS
viii
6.3
Vector potential
293
6.4
Field of any current-carrying wire
296
6.5
Fields of rings and coils
299
6.6
Change in B at a current sheet
303
6.7
Flow the fields transform
306
6.8
Rowland’s experiment
314
6.9
Electrical conduction in a magnetic field:
the Flail effect
314
6.10
Applications
317
Chapter summary
322
Problems
323
Exercises
331
CHAPTER 7
ELECTROMAGNETIC INDUCTION
342
7.1
Faraday’s discovery
343
7.2
Conducting rod moving through a uniform
magnetic field
345
7.3
Loop moving through a nonuniform magnetic field
346
7.4
Stationary loop with the field source moving
352
7.5
Universal law of induction
355
7.6
Mutual inductance
359
7.7
A reciprocity theorem
362
7.8
Self-inductance
364
7.9
Circuit containing self-inductance
366
7.10
Energy stored in the magnetic field
368
7.11
Applications
369
Chapter summary
373
Problems
374
Exercises
380
CHAPTER 8
ALTERNATING-CURRENT CIRCUITS
388
8.1
A resonant circuit
388
8.2
Alternating current
394
8.3
Complex exponential solutions
402
8.4
Alternating-current networks
405
8.5
Admittance and impedance
408
8.6
Power and energy in alternating-current circuits
415
8.7
Applications
418
Chapter summary
420
Problems
421
Exercises
424
CONTENTS
IX
CHAPTER 9
MAXWELL’S EQUATIONS AND ELECTROMAGNETIC
WAVES
430
9.1
“Something is missing”
430
9.2
The displacement current
433
9.3
Maxwell’s equations
436
9.4
An electromagnetic wave
438
9.5
Other waveforms; superposition of waves
441
9.6
Energy transport by electromagnetic waves
446
9.7
How a wave looks in a different frame
452
9.8
Applications
454
Chapter summary
455
Problems
457
Exercises
461
CHAPTER 10
ELECTRIC FIELDS IN MATTER
466
10.1
Dielectrics
467
10.2
The moments of a charge distribution
471
10.3
The potential and field of a dipole
474
10.4
The torque and the force on a dipole in an
external field
477
10.5
Atomic and molecular dipoles; induced
dipole moments
479
10.6
Permanent dipole moments
482
10.7
The electric field caused by polarized matter
483
10.8
Another look at the capacitor
489
10.9
The field of a polarized sphere
492
10.10
A dielectric sphere in a uniform field
495
10.11
The field of a charge in a dielectric medium, and
Gauss’s law
497
10.12
A microscopic view of the dielectric
500
10.13
Polarization in changing fields
504
10.14
The bound-charge current
505
10.15
An electromagnetic wave in a dielectric
507
10.16
Applications
509
Chapter summary
511
Problems
513
Exercises
516
CHAPTER 11
MAGNETIC FIELDS IN MATTER
523
11.1
How various substances respond to a
magnetic field
524
X
CONTENTS
11.2 The absence of magnetic “charge” 529
11.3 The field of a current loop 531
11.4 The force on a dipole in an external field 535
11.5 Electric currents in atoms 540
11.6 Electron spin and magnetic moment 546
11.7 Magnetic susceptibility 549
11.8 The magnetic field caused by magnetized matter 551
11.9 The field of a permanent magnet 557
11.10 Free currents, and the field H 559
11.11 Ferromagnetism 565
11.12 Applications 570
Chapter summary 573
Problems 575
Exercises 577
CHAPTER 12
SOLUTIONS TO THE PROBLEMS 586
12.1 Chapter 1 586
12.2 Chapter 2 611
12.3 Chapter 3 636
12.4 Chapter 4 660
12.5 Chapter 5 678
12.6 Chapter 6 684
12.7 Chapter 7 707
12.8 Chapter 8 722
12.9 Chapter 9 734
12.10 Chapter 10 744
12.11 Chapter 11 755
Appendix A:
Differences between SI and Gaussian units 762
Appendix B:
SI units of common quantities 769
Appendix C:
Unit conversions 774
Appendix D:
SI and Gaussian formulas 778
Appendix E:
Exact relations among SI and Gaussian units 789
CONTENTS
XI
Appendix F:
Curvilinear coordinates
791
Appendix G:
A short review of special relativity
804
Appendix H:
Radiation by an accelerated charge
812
Appendix 1:
Superconductivity
817
Appendix J:
Magnetic resonance
821
Appendix K:
Helpful formulas/facts
825
References
831
Index
833
For 50 years, physics students have enjoyed learning about electricity
and magnetism through the first two editions of this book. The purpose
of the present edition is to bring certain things up to date and to add new
material, in the hopes that the trend will continue. The main changes
from the second edition are (1) the conversion from Gaussian units to SI
units, and (2) the addition of many solved problems and examples.
The first of these changes is due to the fact that the vast majority
of courses on electricity and magnetism are now taught in SI units. The
second edition fell out of print at one point, and it was hard to watch such
a wonderful book fade away because it wasn’t compatible with the way
the subject is presently taught. Of course, there are differing opinions as
to which system of units is “better” for an introductory course. But this
issue is moot, given the reality of these courses.
For students interested in working with Gaussian units, or for instruc¬
tors who want their students to gain exposure to both systems, I have
created a number of appendices that should be helpful. Appendix A dis¬
cusses the differences between the SI and Gaussian systems. Appendix C
derives the conversion factors between the corresponding units in the
two systems. Appendix D explains how to convert formulas from SI to
Gaussian; it then lists, side by side, the SI and Gaussian expressions for
every important result in the book. A little time spent looking at this
appendix will make it clear how to convert formulas from one system to
the other.
The second main change in the book is the addition of many solved
problems, and also many new examples in the text. Each chapter ends
with “problems” and “exercises.” The solutions to the “problems” are
located in Chapter 12. The only official difference between the problems
Preface to the third
edition of Volume 2
XIV
Preface to the third edition of Volume 2
and exercises is that the problems have solutions included, whereas the
exercises do not. (A separate solutions manual for the exercises is avail¬
able to instructors.) In practice, however, one difference is that some of
the more theorem-ish results are presented in the problems, so that stu¬
dents can use these results in other problems/exercises.
Some advice on using the solutions to the problems: problems (and
exercises) are given a (very subjective) difficulty rating from 1 star to 4
stars. If you are having trouble solving a problem, it is critical that you
don’t look at the solution too soon. Brood over it for a while. If you do
finally look at the solution, don’t just read it through. Instead, cover it up
with a piece of paper and read one line at a time until you reach a hint
to get you started. Then set the book aside and work things out for real.
That’s the only way it will sink in. It’s quite astonishing how unhelpful
it is simply to read a solution. You’d think it would do some good, but
in fact it is completely ineffective in raising your understanding to the
next level. Of course, a careful reading of the text, including perhaps a
few problem solutions, is necessary to get the basics down. But if Level
1 is understanding the basic concepts, and Level 2 is being able to apply
those concepts, then you can read and read until the cows come home,
and you’ll never get past Level 1.
The overall structure of the text is essentially the same as in the sec¬
ond edition, although a few new sections have been added. Section 2.7
introduces dipoles. The more formal treatment of dipoles, along with
their applications, remains in place in Chapter 10. But because the funda¬
mentals of dipoles can be understood using only the concepts developed
in Chapters 1 and 2, it seems appropriate to cover this subject earlier
in the book. Section 8.3 introduces the important technique of solving
differential equations by forming complex solutions and then taking the
real part. Section 9.6.2 deals with the Poynting vector, which opens up
the door to some very cool problems.
Each chapter concludes with a list of “everyday” applications of
electricity and magnetism. The discussions are brief. The main purpose
of these sections is to present a list of fun topics that deserve further
investigation. You can carry onward with some combination of books/
internet/people/pondering. There is effectively an infinite amount of in¬
formation out there (see the references at the beginning of Section 1.16
for some starting points), so my goal in these sections is simply to pro¬
vide a springboard for further study.
The intertwined nature of electricity, magnetism, and relativity is
discussed in detail in Chapter 5. Many students find this material highly
illuminating, although some find it a bit difficult. (However, these two
groups are by no means mutually exclusive!) For instructors who wish to
take a less theoretical route, it is possible to skip directly from Chapter 4
to Chapter 6, with only a brief mention of the main result from Chapter 5,
namely the magnetic field due to a straight current-carrying wire.
Preface to the third edition of Volume 2
xv
The use of non-Cartesian coordinates (cylindrical, spherical) is more
prominent in the present edition. For setups possessing certain symme¬
tries, a wisely chosen system of coordinates can greatly simplify the cal¬
culations. Appendix F gives a review of the various vector operators in
the different systems.
Compared with the second edition, the level of difficulty of the
present edition is slightly higher, due to a number of hefty problems that
have been added. If you are looking for an extra challenge, these prob¬
lems should keep you on your toes. Flowever, if these are ignored (which
they certainly can be, in any standard course using this book), then the
level of difficulty is roughly the same.
I am grateful to all the students who used a draft version of this book
and provided feedback. Their input has been invaluable. I would also like
to thank Jacob Barandes for many illuminating discussions of the more
subtle topics in the book. Paul Horowitz helped get the project off the
ground and has been an endless supplier of cool facts. It was a plea¬
sure brainstorming with Andrew Milewski, who offered many ideas for
clever new problems. Howard Georgi and Wolfgang Rueckner provided
much-appreciated sounding boards and sanity checks. Takuya Kitagawa
carefully read through a draft version and offered many helpful sug¬
gestions. Other friends and colleagues whose input I am grateful for
are: Allen Crockett, David Derbes, John Doyle, Gary Feldman, Melissa
Franklin, Jerome Fung, Jene Golovchenko, Doug Goodale, Robert Hart,
Tom Hayes, Peter Hedman, Jennifer Hoffman, Charlie Holbrow, Gareth
Kafka, Alan Levine, Aneesh Manohar, Kirk McDonald, Masahiro Morii,
Lev Okun, Joon Pahk, Dave Patterson, Mara Prentiss, Dennis Purcell,
Frank Purcell, Daniel Rosenberg, Emily Russell, Roy Shwitters, Nils
Sorensen, Josh Winn, and Amir Yacoby.
I would also like to thank the editorial and production group at Cam¬
bridge University Press for their professional work in transforming the
second edition of this book into the present one. It has been a pleasure
working with Lindsay Barnes, Simon Capelin, Irene Pizzie, Charlotte
Thomas, and Ali Woollatt.
Despite careful editing, there is zero probability that this book is
error free. A great deal of new material has been added, and errors have
undoubtedly crept in. If anything looks amiss, please check the webpage
www.cambridge.org/Purcell-Morin for a list of typos, updates, etc. And
please let me know if you discover something that isn’t already posted.
Suggestions are always welcome.
David Morin
This revision of “Electricity and Magnetism,” Volume 2 of the Berkeley
Physics Course, has been made with three broad aims in mind. First, I
have tried to make the text clearer at many points. In years of use teachers
and students have found innumerable places where a simplification or
reorganization of an explanation could make it easier to follow. Doubtless
some opportunities for such improvements have still been missed; not too
many, I hope.
A second aim was to make the book practically independent of its
companion volumes in the Berkeley Physics Course. As originally con¬
ceived it was bracketed between Volume I, which provided the needed
special relativity, and Volume 3, “Waves and Oscillations,” to which
was allocated the topic of electromagnetic waves. As it has turned out,
Volume 2 has been rather widely used alone. In recognition of that I have
made certain changes and additions. A concise review of the relations of
special relativity is included as Appendix A. Some previous introduction
to relativity is still assumed. The review provides a handy reference and
summary for the ideas and formulas we need to understand the fields of
moving charges and their transformation from one frame to another. The
development of Maxwell’s equations for the vacuum has been transferred
from the heavily loaded Chapter 7 (on induction) to a new Chapter 9,
where it leads naturally into an elementary treatment of plane electro¬
magnetic waves, both running and standing. The propagation of a wave
in a dielectric medium can then be treated in Chapter 10 on Electric
Fields in Matter.
A third need, to modernize the treatment of certain topics, was most
urgent in the chapter on electrical conduction. A substantially rewritten
Preface to the
second edition of
Volume 2
Preface to the second edition of Volume 2
xviii
Chapter 4 now includes a section on the physics of homogeneous semi¬
conductors, including doped semiconductors. Devices are not included,
not even a rectifying junction, but what is said about bands, and donors
and acceptors, could serve as starting point for development of such top¬
ics by the instructor. Thanks to solid-state electronics the physics of the
voltaic cell has become even more relevant to daily life as the number
of batteries in use approaches in order of magnitude the world’s popu¬
lation. In the first edition of this book I unwisely chose as the example
of an electrolytic cell the one cell—the Weston standard cell—which
advances in physics were soon to render utterly obsolete. That section
has been replaced by an analysis, with new diagrams, of the lead-acid
storage battery—ancient, ubiquitous, and far from obsolete.
One would hardly have expected that, in the revision of an elemen¬
tary text in classical electromagnetism, attention would have to be paid to
new developments in particle physics. But that is the case for two ques¬
tions that were discussed in the first edition, the significance of charge
quantization, and the apparent absence of magnetic monopoles. Obser¬
vation of proton decay would profoundly affect our view of the first ques¬
tion. Assiduous searches for that, and also for magnetic monopoles, have
at this writing yielded no confirmed events, but the possibility of such
fundamental discoveries remains open.
Three special topics, optional extensions of the text, are introduced
in short appendixes: Appendix B: Radiation by an Accelerated Charge;
Appendix C: Superconductivity; and Appendix D: Magnetic Resonance.
Our primary system of units remains the Gaussian CGS system. The
SI units, ampere, coulomb, volt, ohm, and tesla are also introduced in
the text and used in many of the problems. Major formulas are repeated
in their SI formulation with explicit directions about units and conver¬
sion factors. The charts inside the back cover summarize the basic rela¬
tions in both systems of units. A special chart in Chapter 11 reviews, in
both systems, the relations involving magnetic polarization. The student
is not expected, or encouraged, to memorize conversion factors, though
some may become more or less familiar through use, but to look them up
whenever needed. There is no objection to a “mixed” unit like the ohm-
cm, still often used for resistivity, providing its meaning is perfectly clear.
The definition of the meter in terms of an assigned value for the
speed of light, which has just become official, simplifies the exact rela¬
tions among the units, as briefly explained in Appendix E.
There are some 300 problems, more than half of them new.
It is not possible to thank individually all the teachers and students
who have made good suggestions for changes and corrections. I fear
that some will be disappointed to find that their suggestions have not
been followed quite as they intended. That the net result is a substantial
improvement I hope most readers familiar with the first edition will agree.
Preface to the second edition of Volume 2
XIX
Mistakes both old and new will surely be found. Communications pointing
them out will be gratefully received.
It is a pleasure to thank Olive S. Rand for her patient and skillful
assistance in the production of the manuscript.
Edward M. Purcell
The subject of this volume of the Berkeley Physics Course is electricity
and magnetism. The sequence of topics, in rough outline, is not unusual:
electrostatics; steady currents; magnetic field; electromagnetic induc¬
tion; electric and magnetic polarization in matter. However, our approach
is different from the traditional one. The difference is most conspicu¬
ous in Chaps. 5 and 6 where, building on the work of Vol. I, we treat
the electric and magnetic fields of moving charges as manifestations of
relativity and the invariance of electric charge. This approach focuses
attention on some fundamental questions, such as: charge conservation,
charge invariance, the meaning of field. The only formal apparatus of
special relativity that is really necessary is the Lorentz transformation
of coordinates and the velocity-addition formula. It is essential, though,
that the student bring to this part of the course some of the ideas and atti¬
tudes Vol. I sought to develop—among them a readiness to look at things
from different frames of reference, an appreciation of invariance, and a
respect for symmetry arguments. We make much use also, in Vol. II, of
arguments based on superposition.
Our approach to electric and magnetic phenomena in matter is pri¬
marily “microscopic,” with emphasis on the nature of atomic and molec¬
ular dipoles, both electric and magnetic. Electric conduction, also, is
described microscopically in the terms of a Drude-Lorentz model. Nat¬
urally some questions have to be left open until the student takes up
quantum physics in Vol. IV. But we freely talk in a matter-of-fact way
about molecules and atoms as electrical structures with size, shape, and
stiffness, about electron orbits, and spin. We try to treat carefully a ques¬
tion that is sometimes avoided and sometimes beclouded in introductory
texts, the meaning of the macroscopic fields E and B inside a material.
Preface to the first
edition of Volume 2
XXII
Preface to the first edition of Volume 2
In Vol. II, the student’s mathematical equipment is extended by
adding some tools of the vector calculus—gradient, divergence, curl,
and the Laplacian. These concepts are developed as needed in the early
chapters.
In its preliminary versions, Vol. II has been used in several classes at
the University of California. It has benefited from criticism by many peo¬
ple connected with the Berkeley Course, especially from contributions
by E. D. Commins and F. S. Crawford, Jr., who taught the first classes to
use the text. They and their students discovered numerous places where
clarification, or something more drastic, was needed; many of the revi¬
sions were based on their suggestions. Students’ criticisms of the last
preliminary version were collected by Robert Goren, who also helped
to organize the problems. Valuable criticism has come also from J. D.
Gavenda, who used the preliminary version at the University of Texas,
and from E. F. Taylor, of Wesleyan University. Ideas were contributed by
Allan Kaufman at an early stage of the writing. A. Felzer worked through
most of the first draft as our first “test student.”
The development of this approach to electricity and magnetism was
encouraged, not only by our original Course Committee, but by col¬
leagues active in a rather parallel development of new course material
at the Massachusetts Institute of Technology. Among the latter, J. R.
Tessman, of the MIT Science Teaching Center and Tufts University, was
especially helpful and influential in the early formulation of the strategy.
He has used the preliminary version in class, at MIT, and his critical
reading of the entire text has resulted in many further changes and cor¬
rections.
Publication of the preliminary version, with its successive revisions,
was supervised by Mrs. Mary R. Maloney. Mrs. Lila Lowell typed most
of the manuscript. The illustrations were put into final form by Felix
Cooper.
The author of this volume remains deeply grateful to his friends
in Berkeley, and most of all to Charles Kittel, for the stimulation and
constant encouragement that have made the long task enjoyable.
Edward M. Purcell
Overview The existence of this book is owed (both figuratively
and literally) to the fact that the building blocks of matter possess a
quality called charge. Two important aspects of charge are conser¬
vation and quantization. The electric force between two charges
is given by Coulomb’s law. Like the gravitational force, the electric
force falls off like 1/r 2 . It is conservative, so we can talk about the
potential energy of a system of charges (the work done in assem¬
bling them). A very useful concept is the electric field, which is
defined as the force per unit charge. Every point in space has a
unique electric field associated with it. We can define the flux of
the electric field through a given surface. This leads us to Gauss’s
law, which is an alternative way of stating Coulomb’s law. In cases
involving sufficient symmetry, it is much quicker to calculate the
electric field via Gauss’s law than via Coulomb’s law and direct
integration. Finally, we discuss the energy density in the elec¬
tric field, which provides another way of calculating the potential
energy of a system.
Electrostatics:
charges and fields
1.1 Electric charge
Electricity appeared to its early investigators as an extraordinary phe¬
nomenon. To draw from bodies the “subtle fire,” as it was sometimes
called, to bring an object into a highly electrified state, to produce a
steady flow of current, called for skillful contrivance. Except for the
spectacle of lightning, the ordinary manifestations of nature, from the
freezing of water to the growth of a tree, seemed to have no relation to
the curious behavior of electrified objects. We know now that electrical
2
Electrostatics: charges and fields
forces largely determine the physical and chemical properties of matter
over the whole range from atom to living cell. For this understanding we
have to thank the scientists of the nineteenth century. Ampere, Faraday,
Maxwell, and many others, who discovered the nature of electromag¬
netism, as well as the physicists and chemists of the twentieth century
who unraveled the atomic structure of matter.
Classical electromagnetism deals with electric charges and currents
and their interactions as if all the quantities involved could be measured
independently, with unlimited precision. Here classical means simply
“nonquantum.” The quantum law with its constant h is ignored in the
classical theory of electromagnetism, just as it is in ordinary mechanics.
Indeed, the classical theory was brought very nearly to its present state
of completion before Planck’s discovery of quantum effects in 1900. It
has survived remarkably well. Neither the revolution of quantum physics
nor the development of special relativity dimmed the luster of the elec¬
tromagnetic field equations Maxwell wrote down 150 years ago.
Of course the theory was solidly based on experiment, and because
of that was fairly secure within its original range of application - to
coils, capacitors, oscillating currents, and eventually radio waves and
light waves. But even so great a success does not guarantee validity in
another domain, for instance, the inside of a molecule.
Two facts help to explain the continuing importance in modern
physics of the classical description of electromagnetism. First, special
relativity required no revision of classical electromagnetism. Historic¬
ally speaking, special relativity grew out of classical electromagnetic
theory and experiments inspired by it. Maxwell’s field equations, devel¬
oped long before the work of Lorentz and Einstein, proved to be entirely
compatible with relativity. Second, quantum modifications of the elec¬
tromagnetic forces have turned out to be unimportant down to distances
less than 10“ 12 meters, 100 times smaller than the atom. We can describe
the repulsion and attraction of particles in the atom using the same laws
that apply to the leaves of an electroscope, although we need quantum
mechanics to predict how the particles will behave under those forces.
For still smaller distances, a fusion of electromagnetic theory and quan¬
tum theory, called quantum electrodynamics, has been remarkably suc¬
cessful. Its predictions are confirmed by experiment down to the smallest
distances yet explored.
It is assumed that the reader has some acquaintance with the elemen¬
tary facts of electricity. We are not going to review all the experiments
by which the existence of electric charge was demonstrated, nor shall we
review all the evidence for the electrical constitution of matter. On the
other hand, we do want to look carefully at the experimental foundations
of the basic laws on which all else depends. In this chapter we shall study
the physics of stationary electric charges - electrostatics.
Certainly one fundamental property of electric charge is its exis¬
tence in the two varieties that were long ago named positive and negative.
1.1 Electric charge
3
The observed fact is that all charged particles can be divided into two
classes such that all members of one class repel each other, while attract¬
ing members of the other class. If two small electrically charged bodies
A and B, some distance apart, attract one another, and if A attracts some
third electrified body C, then we always find that B repels C. Contrast
this with gravitation: there is only one kind of gravitational mass, and
every mass attracts every other mass.
One may regard the two kinds of charge, positive and negative, as
opposite manifestations of one quality, much as right and left are the
two kinds of handedness. Indeed, in the physics of elementary parti¬
cles, questions involving the sign of the charge are sometimes linked to a
question of handedness, and to another basic symmetry, the relation of a
sequence of events, a , then b, then c, to the temporally reversed sequence
c, then b, then a. It is only the duality of electric charge that concerns us
here. For every kind of particle in nature, as far as we know, there can
exist an antiparticle, a sort of electrical “mirror image.” The antiparticle
carries charge of the opposite sign. If any other intrinsic quality of the
particle has an opposite, the antiparticle has that too, whereas in a prop¬
erty that admits no opposite, such as mass, the antiparticle and particle
are exactly alike.
The electron’s charge is negative; its antiparticle, called a positron,
has a positive charge, but its mass is precisely the same as that of the
electron. The proton’s antiparticle is called simply an antiproton; its elec¬
tric charge is negative. An electron and a proton combine to make an
ordinary hydrogen atom. A positron and an antiproton could combine
in the same way to make an atom of antihydrogen. Given the building
blocks, positrons, antiprotons, and antineutrons, there could be built
up the whole range of antimatter, from antihydrogen to antigalaxies.
There is a practical difficulty, of course. Should a positron meet an elec¬
tron or an antiproton meet a proton, that pair of particles will quickly
vanish in a burst of radiation. It is therefore not surprising that even
positrons and antiprotons, not to speak of antiatoms, are exceedingly
rare and short-lived in our world. Perhaps the universe contains, some¬
where, a vast concentration of antimatter. If so, its whereabouts is a
cosmological mystery.
The universe around us consists overwhelmingly of matter, not anti¬
matter. That is to say, the abundant carriers of negative charge are
electrons, and the abundant carriers of positive charge are protons. The
proton is nearly 2000 times heavier than the electron, and very different,
too, in some other respects. Thus matter at the atomic level incorpo¬
rates negative and positive electricity in quite different ways. The posi¬
tive charge is all in the atomic nucleus, bound within a massive structure
no more than 10 -14 m in size, while the negative charge is spread, in
1 Although the electric charge of each is zero, the neutron and its antiparticle are not
interchangeable. In certain properties that do not concern us here, they are opposite.
4
Electrostatics: charges and fields
effect, through a region about 10 4 times larger in dimensions. It is hard
to imagine what atoms and molecules - and all of chemistry - would be
like, if not for this fundamental electrical asymmetry of matter.
What we call negative charge, by the way, could just as well have
been called positive. The name was a historical accident. There is nothing
essentially negative about the charge of an electron. It is not like a neg¬
ative integer. A negative integer, once multiplication has been defined,
differs essentially from a positive integer in that its square is an integer
of opposite sign. But the product of two charges is not a charge; there is
no comparison.
Two other properties of electric charge are essential in the electrical
structure of matter: charge is conserved, and charge is quantized. These
properties involve quantity of charge and thus imply a measurement of
charge. Presently we shall state precisely how charge can be measured in
terms of the force between charges a certain distance apart, and so on.
But let us take this for granted for the time being, so that we may talk
freely about these fundamental facts.
Figure 1.1.
Charged particles are created in pairs with
equal and opposite charge.
1.2 Conservation of charge
The total charge in an isolated system never changes. By isolated we
mean that no matter is allowed to cross the boundary of the system. We
could let light pass into or out of the system, since the “particles” of light,
called photons, carry no charge at all. Within the system charged parti¬
cles may vanish or reappear, but they always do so in pairs of equal and
opposite charge. For instance, a thin-walled box in a vacuum exposed to
gamma rays might become the scene of a “pair-creation” event in which
a high-energy photon ends its existence with the creation of an electron
and a positron (Fig. 1.1). Two electrically charged particles have been
newly created, but the net change in total charge, in and on the box, is
zero. An event that would violate the law we have just stated would be
the creation of a positively charged particle without the simultaneous cre¬
ation of a negatively charged particle. Such an occurrence has never been
observed.
Of course, if the electric charges of an electron and a positron were
not precisely equal in magnitude, pair creation would still violate the
strict law of charge conservation. That equality is a manifestation of the
particle-antiparticle duality already mentioned, a universal symmetry of
nature.
One thing will become clear in the course of our study of electro¬
magnetism: nonconservation of charge would be quite incompatible with
the structure of our present electromagnetic theory. We may therefore
state, either as a postulate of the theory or as an empirical law supported
without exception by all observations so far, the charge conservation law:
1.3 Quantization of charge
5
The total electric charge in an isolated system, that is, the algebraic
sum of the positive and negative charge present at any time, never
changes.
Sooner or later we must ask whether this law meets the test of rel¬
ativistic invariance. We shall postpone until Chapter 5 a thorough dis¬
cussion of this important question. But the answer is that it does, and
not merely in the sense that the statement above holds in any given iner¬
tial frame, but in the stronger sense that observers in different frames,
measuring the charge, obtain the same number. In other words, the total
electric charge of an isolated system is a relativistically invariant number.
1.3 Quantization of charge
The electric charges we find in nature come in units of one magnitude
only, equal to the amount of charge carried by a single electron. We
denote the magnitude of that charge by e. (When we are paying atten¬
tion to sign, we write — e for the charge on the electron itself.) We have
already noted that the positron carries precisely that amount of charge,
as it must if charge is to be conserved when an electron and a positron
annihilate, leaving nothing but light. What seems more remarkable is the
apparently exact equality of the charges carried by all other charged par¬
ticles - the equality, for instance, of the positive charge on the proton and
the negative charge on the electron.
That particular equality is easy to test experimentally. We can see
whether the net electric charge carried by a hydrogen molecule, which
consists of two protons and two electrons, is zero. In an experiment
carried out by J. G. King,- hydrogen gas was compressed into a tank
that was electrically insulated from its surroundings. The tank contained
about 5 • 10 24 molecules (approximately 17 grams) of hydrogen. The gas
was then allowed to escape by means that prevented the escape of any
ion - a molecule with an electron missing or an extra electron attached.
If the charge on the proton differed from that on the electron by, say, one
part in a billion, then each hydrogen molecule would carry a charge of
2 ■ 10 -9 e, and the departure of the whole mass of hydrogen would alter
the charge of the tank by 10 16 e, a gigantic effect. In fact, the experiment
could have revealed a residual molecular charge as small as 2 • 10 _20 e,
and none was observed. This proved that the proton and the electron do
not differ in magnitude of charge by more than 1 part in 10 2 °.
Perhaps the equality is really exact for some reason we don’t yet
understand. It may be connected with the possibility, suggested by certain
" See King (1960). References to previous tests of charge equality will be found in this
article and in the chapter by V. W. Hughes in Hughes (1964).
6
Electrostatics: charges and fields
theories, that a proton can, very rarely, decay into a positron and some
uncharged particles. If that were to occur, even the slightest discrepancy
between proton charge and positron charge would violate charge conser¬
vation. Several experiments designed to detect the decay of a proton have
not yet, as of this writing, registered with certainty a single decay. If and
when such an event is observed, it will show that exact equality of the
magnitude of the charge of the proton and the charge of the electron (the
positron’s antiparticle) can be regarded as a corollary of the more general
law of charge conservation.
That notwithstanding, we now know that the internal structure of all
the strongly interacting particles called hadrons - a class that includes
the proton and the neutron - involves basic units called quarks, whose
electric charges come in multiples of e/3. The proton, for example, is
made with three quarks, two with charge 2e/3 and one with charge — e/3.
The neutron contains one quark with charge 2e/3 and two quarks with
charge —e/3.
Several experimenters have searched for single quarks, either free or
attached to ordinary matter. The fractional charge of such a quark, since
it cannot be neutralized by any number of electrons or protons, should
betray the quark’s presence. So far no fractionally charged particle has
been conclusively identified. The present theory of the strong interac¬
tions, called quantum chromodynamics, explains why the liberation of a
quark from a hadron is most likely impossible.
The fact of charge quantization lies outside the scope of classical
electromagnetism, of course. We shall usually ignore it and act as if our
point charges q could have any strength whatsoever. This will not get us
into trouble. Still, it is worth remembering that classical theory cannot
be expected to explain the structure of the elementary particles. (It is not
certain that present quantum theory can either!) What holds the electron
together is as mysterious as what fixes the precise value of its charge.
Something more than electrical forces must be involved, for the electro¬
static forces between different parts of the electron would be repulsive.
In our study of electricity and magnetism we shall treat the charged
particles simply as carriers of charge, with dimensions so small that
their extension and structure is, for most purposes, quite insignificant.
In the case of the proton, for example, we know from high-energy scat¬
tering experiments that the electric charge does not extend appreciably
beyond a radius of 10“ 15 m. We recall that Rutherford’s analysis of the
scattering of alpha particles showed that even heavy nuclei have their
electric charge distributed over a region smaller than 10“ 13 m. For the
physicist of the nineteenth century a “point charge” remained an abstract
notion. Today we are on familiar terms with the atomic particles. The
graininess of electricity is so conspicuous in our modern description of
nature that we find a point charge less of an artificial idealization than a
smoothly varying distribution of charge density. When we postulate such
smooth charge distributions, we may think of them as averages over very
1.4 Coulomb’s law
7
large numbers of elementary charges, in the same way that we can define
the macroscopic density of a liquid, its lumpiness on a molecular scale
notwithstanding.
1.4 Coulomb’s law
As you probably already know, the interaction between electric charges
at rest is described by Coulomb’s law: two stationary electric charges
repel or attract one another with a force proportional to the product of
the magnitude of the charges and inversely proportional to the square of
the distance between them.
We can state this compactly in vector form:
F 2 =k
qiqirn
(l.l)
'21
Here q\ and qi are numbers (scalars) giving the magnitude and sign of
the respective charges, r 2 i is the unit vector in the direction from charge
1 to charge 2, and F 2 is the force acting on charge 2. Thus Eq. (1.1)
expresses, among other things, the fact that like charges repel and unlike
charges attract. Also, the force obeys Newton’s third law; that is,
F 2 = — F[.
The unit vector r 2 i shows that the force is parallel to the line joining
the charges. It could not be otherwise unless space itself has some built-
in directional property, for with two point charges alone in empty and
isotropic space, no other direction could be singled out.
If the point charge itself had some internal structure, with an axis
defining a direction, then it would have to be described by more than the
mere scalar quantity q. It is true that some elementary particles, includ¬
ing the electron, do have another property, called spin. This gives rise to
a magnetic force between two electrons in addition to their electrostatic
repulsion. This magnetic force does not, in general, act in the direction
of the line joining the two particles. It decreases with the inverse fourth
power of the distance, and at atomic distances of 10 -10 m the Coulomb
force is already about 10 4 times stronger than the magnetic interaction
of the spins. Another magnetic force appears if our charges are moving -
hence the restriction to stationary charges in our statement of Coulomb’s
law. We shall return to these magnetic phenomena in later chapters.
Of course we must assume, in writing Eq. (1.1), that both charges
are well localized, each occupying a region small compared with r 2 i.
Otherwise we could not even define the distance r 2 i precisely.
The value of the constant k in Eq. (1.1) depends on the units in which
r, F, and q are to be expressed. In this book we will use the International
System of Units, or “SI” units for short. This system is based on the
3
The convention we adopt here may not seem the natural choice, but it is more
consistent with the usage in some other parts of physics and we shall try to follow it
throughout this book.
8
Electrostatics: charges and fields
1 newton = 10 5 dynes
1 coulomb = 2.998 X 10 9 esu
e = 4.802 X 1(T 10 esu = 1.602 X 10~ 19 coulomb
F = 8.988 x 10 s newtons
2 coulombs.
3ulombs^^^^
or
5 coulombs
5 coulombs
F =
/
newtons
coulomb
/I
F = 8.988 x 10 8 newtons 8 - 988 x 1()9
6 0 = 8.854 x 10- 12
Figure 1.2.
Coulomb’s law expressed in Gaussian
electrostatic units (top) and in SI units (bottom).
The constant e 0 and the factor relating coulombs
to esu are connected, as we shall learn later,
with the speed of light. We have rounded off the
constants in the figure to four-digit accuracy.
The precise values are given in Appendix E.
meter, kilogram, and second as units of length, mass, and time. The SI
unit of charge is the coulomb (C). Some other SI electrical units that
we will eventually become familiar with are the volt, ohm, ampere, and
tesla. The official definition of the coulomb involves the magnetic force,
which we will discuss in Chapter 6. For present purposes, we can define
the coulomb as follows. Two like charges, each of 1 coulomb, repel one
another with a force of 8.988 ■ 10 9 newtons when they are 1 meter apart.
In other words, the k in Eq. (1.1) is given by
q Nm 2
k = 8.988 ■ 10 9 —g2~- (1.2)
In Chapter 6 we will learn where this seemingly arbitrary value of k
comes from. In general, approximating k by 9 ■ 10 9 N m 2 /C 2 is quite suf¬
ficient. The magnitude of e, the fundamental quantum of electric charge,
happens to be about 1.602 ■ 10“ 19 C. So if you wish, you may think of
a coulomb as defined to be the magnitude of the charge contained in
6.242 • 10 18 electrons.
Instead of k, it is customary (for historical reasons) to introduce a
constant €q which is defined by
47T60
(1.3)
C 2 s 2
kgm 3
In terms of eo, Coulomb’s law in Eq. (1.1) takes the form
1 qiq 2 r 2 l
47reo r \j
(1.4)
The constant eo will appear in many expressions that we will meet in the
course of our study. The 47r is included in the definition of eg so that
certain formulas (such as Gauss’s law in Sections 1.10 and 2.9) take on
simple forms. Additional details and technicalities concerning eo can be
found in Appendix E.
Another system of units that comes up occasionally is the Gaus¬
sian system, which is one of several types of cgs systems, short for
centimeter-gram-second. (In contrast, the SI system is an mks system,
short for meter-kilogram-second.) The Gaussian unit of charge is the
“electrostatic unit,” or esu. The esu is defined so that the constant k
in Eq. (1.1) exactly equals 1 (and this is simply the number 1, with no
units) when r 2 \ is measured in cm, F in dynes, and the q values in esu.
Figure 1.2 gives some examples using the SI and Gaussian systems of
units. Further discussion of the SI and Gaussian systems can be found in
Appendix A.
1.4 Coulomb’s law
9
Example (Relation between 1 coulomb and 1 esu) Show that 1 coulomb
equals 2.998 • 10 9 esu (which generally can be approximated by 3 ■ 10 9 esu).
Solution From Eqs. (1.1) and (1.2), two charges of 1 coulomb separated by a
distance of 1 m exert a (large!) force of 8.988 • 10 9 N ^ 9 - 10 9 N on each other.
We can convert this to the Gaussian unit of force via
. N _ . kgm _ (1000g)(100cm)
s 2 s 2
10 5 dynes.
(1.5)
The two 1 C charges therefore exert a force of 9 • 10 14 dynes on each other. How
would someone working in Gaussian units describe this situation? In Gaussian
units, Coulomb's law gives the force simply as q 2 /r~. The separation is 100 cm,
so if 1 coulomb equals N esu (with N to be determined), the 9 • 10 14 dyne force
between the charges can be expressed as
9• 10 14 dyne = (Nesu) =>. (V * 2 * = 9.io 18 => (V = 3-10 9 . (1.6)
(100 cm) 2
Hence, 4 * * *
1 C = 3 ■ 10 9 esu. (1.7)
The magnitude of the electron charge is then given approximately by e= 1.6-
10“ 19 C 4.8- 10“ 10 esu.
If we had used the more exact value of k in Eq. (1.2), the “3” in our result
would have been replaced by V8.988 = 2.998. This looks suspiciously similar to
the “2.998” in the speed of light, c = 2.998 • 10 8 m/s. This is no coincidence. We
will see in Section 6.1 that Eq. (1.7) can actually be written as 1 C = (10(c)) esu,
where we have put the c in brackets to signify that it is just the number 2.998 • 10 8
without the units of m/s.
On an everyday scale, a coulomb is an extremely large amount of charge,
as evidenced by the fact that if you have two such charges separated by 1 m
(never mind how you would keep each charge from flying apart due to the self
repulsion!), the above force of 9 • 10 9 N between them is about one million tons.
The esu is a much more reasonable unit to use for everyday charges. For example,
the static charge on a balloon that sticks to your hair is on the order of 10 or
100 esu.
The only way we have of detecting and measuring electric charges
is by observing the interaction of charged bodies. One might wonder,
then, how much of the apparent content of Coulomb’s law is really only
definition. As it stands, the significant physical content is the statement
of inverse-square dependence and the implication that electric charge
4 We technically shouldn't be using an “=” sign here, because it suggests that the units of
a coulomb are the same as those of an esu. This is not the case; they are units in
different systems and cannot be expressed in terms of each other. The proper way to
express Eq. (1.7) is to say, “1 C is equivalent to 3 • 10 9 esu.” But we’ll usually just use
the “=” sign, and you’ 11 know what we mean. See Appendix A for further discussion
of this.
10
Electrostatics: charges and fields
(a)
Great _ ^
distance q 3
(b)
^... Great •
^ v 3 distance
•r'
«i
(c)
<?3
\0 C
Figure 1.3.
The force on q\ in (c) is the sum of the forces on
q\ in (a) and (b).
is additive in its effect. To bring out the latter point, we have to con¬
sider more than two charges. After all, if we had only two charges in
the world to experiment with, <71 and < 72 , we could never measure them
separately. We could verify only that F is proportional to 1 /r^. Suppose
we have three bodies carrying charges q\, < 72 , and < 73 . We can measure
the force on q\ when <72 is 10 cm away from q\, with <73 very far away,
as in Fig. 1.3(a). Then we can take qi away, bring <73 into < 72 ’s former
position, and again measure the force on q\. Finally, we can bring <72
and <73 very close together and locate the combination 10 cm from q\.
We find by measurement that the force on <71 is equal to the sum of the
forces previously measured. This is a significant result that could not
have been predicted by logical arguments from symmetry like the one
we used above to show that the force between two point charges had to
be along the line joining them. The force with which two charges interact
is not changed by the presence of a third charge.
No matter how many charges we have in our system. Coulomb’s law
in Eq. (1.4) can be used to calculate the interaction of every pair. This is
the basis of the principle of superposition, which we shall invoke again
and again in our study of electromagnetism. Superposition means com¬
bining two sets of sources into one system by adding the second system
“on top of’ the first without altering the configuration of either one. Our
principle ensures that the force on a charge placed at any point in the
combined system will be the vector sum of the forces that each set of
sources, acting alone, causes to act on a charge at that point. This prin¬
ciple must not be taken lightly for granted. There may well be a domain
of phenomena, involving very small distances or very intense forces,
where superposition no longer holds. Indeed, we know of quantum phe¬
nomena in the electromagnetic field that do represent a failure of super¬
position, seen from the viewpoint of the classical theory.
Thus the physics of electrical interactions comes into full view only
when we have more than two charges. We can go beyond the explicit
statement of Eq. (1.1) and assert that, with the three charges in Fig. 1.3
occupying any positions whatsoever, the force on any one of them, such
as < 73 , is correctly given by the following equation:
1 <73<7lf 31 1 qiqihl ..
“ 47reo r 2 31 4ne 0 r] 2 ' ^ j
The experimental verification of the inverse-square law of electri¬
cal attraction and repulsion has a curious history. Coulomb himself ann¬
ounced the law in 1786 after measuring with a torsion balance the force
between small charged spheres. But 20 years earlier Joseph Priestly, car¬
rying out an experiment suggested to him by Benjamin Franklin, had
noticed the absence of electrical influence within a hollow charged con¬
tainer and made an inspired conjecture: “May we not infer from this
experiment that the attraction of electricity is subject to the same laws
with that of gravitation and is therefore according to the square of the
1.5 Energy of a system of charges
11
distances; since it is easily demonstrated that were the earth in the form
of a shell, a body in the inside of it would not be attracted to one side
more than the other.” (Priestly, 1767).
The same idea was the basis of an elegant experiment in 1772 by
Henry Cavendish. Cavendish charged a spherical conducting shell that
contained within it, and temporarily connected to it, a smaller sphere.
The outer shell was then separated into two halves and carefully removed,
the inner sphere having been first disconnected. This sphere was tested
for charge, the absence of which would confirm the inverse-square law.
(See Problem 2.8 for the theory behind this.) Assuming that a deviation
from the inverse-square law could be expressed as a difference in the
exponent, 2 + 5, say, instead of 2, Cavendish concluded that S must be
less than 0.03. This experiment of Cavendish remained largely unknown
until Maxwell discovered and published Cavendish’s notes a century
later (1876). At that time also. Maxwell repeated the experiment with
improved apparatus, pushing the limit down to 8 < 10 -6 . The present
limit on S is a fantastically small number - about one part in 10 16 ; see
Crandall (1983) and Williams et al. (1971).
Two hundred years after Cavendish’s experiment, however, the ques¬
tion of interest changed somewhat. Never mind how perfectly Coulomb’s
law works for charged objects in the laboratory - is there a range of dis¬
tances where it completely breaks down? There are two domains in either
of which a breakdown is conceivable. The first is the domain of very
small distances, distances less than 10“ 16 m, where electromagnetic the¬
ory as we know it may not work at all. As for very large distances, from
the geographical, say, to the astronomical, a test of Coulomb’s law by
the method of Cavendish is obviously not feasible. Nevertheless we do
observe certain large-scale electromagnetic phenomena that prove that
the laws of classical electromagnetism work over very long distances.
One of the most stringent tests is provided by planetary magnetic fields,
in particular the magnetic field of the giant planet Jupiter, which was
surveyed in the mission of Pioneer 10. The spatial variation of this field
was carefully analyzed^ and found to be entirely consistent with classi¬
cal theory out to a distance of at least 10 5 km from the planet. This is
tantamount to a test, albeit indirect, of Coulomb’s law over that distance.
To summarize, we have every reason for confidence in Coulomb’s
law over the stupendous range of 24 decades in distance, from 10“ 16 to
10 8 m, if not farther, and we take it as the foundation of our description
of electromagnetism.
1.5 Energy of a system of charges
In principle. Coulomb’s law is all there is to electrostatics. Given the
charges and their locations, we can find all the electrical forces. Or, given
5 See Davis el al. (1975). For a review of the history of the exploration of the outer limit
of classical electromagnetism, see Goldhaber and Nieto (1971).
12
Electrostatics: charges and fields
(a) •
<h
Great
distance
(b)
Figure 1.4.
Three charges are brought near one another.
First q 2 is brought in; then, with q\ and q 2 fixed,
q-j is brought in.
that the charges are free to move under the influence of other kinds of
forces as well, we can find the equilibrium arrangement in which the
charge distribution will remain stationary. In the same sense, Newton’s
laws of motion are all there is to mechanics. But in both mechanics and
electromagnetism we gain power and insight by introducing other con¬
cepts, most notably that of energy.
Energy is a useful concept here because electrical forces are con¬
servative. When you push charges around in electric fields, no energy is
irrecoverably lost. Everything is perfectly reversible. Consider first the
work that must be done on the system to bring some charged bodies into
a particular arrangement. Let us start with two charged bodies or parti¬
cles very far apart from one another, as indicated in Fig. 1.4(a), carrying
charges q\ and < 72 . Whatever energy may have been needed to create
these two concentrations of charge originally we shall leave entirely out
of account. How much work does it take to bring the particles slowly
together until the distance between them is r\ 2 ?
It makes no difference whether we bring q\ toward <72 or the other
way around. In either case the work done is the integral of the product:
force times displacement, where these are signed quantities. The force
that has to be applied to move one charge toward the other is equal and
opposite to the Coulomb force. Therefore,
W = J (applied force) • (displacement)
= r a 1 gi r V= 1 qm . d.9)
Jr= OO V 4jr< n r z ) 4ite 0 r\ 2
Note that because r is changing from 00 to ri 2 , the differential dr is
negative. We know that the overall sign of the result is correct, because
the work done on the system must be positive for charges of like sign;
they have to be pushed together (consistent with the minus sign in the
applied force). Both the displacement and the applied force are negative
in this case, resulting in positive work being done on the system. With q\
and <72 in coulombs, and r 12 in meters, Eq. (1.9) gives the work in joules.
This work is the same whatever the path of approach. Let’s review
the argument as it applies to the two charges q\ and <72 in Fig. 1.5. There
we have kept q\ fixed, and we show <72 moved to the same final posi¬
tion along two different paths. Every spherical shell, such as the one
indicated between r and r + dr, must be crossed by both paths. The
increment of work involved, — F • ds in this bit of path (where F is the
Coulomb force), is the same for the two paths .' 1 The reason is that F has
the same magnitude at both places and is directed radially from < 71 , while
6 Here we use for the first time the scalar product, or “dot product,” of two vectors.
A reminder: the scalar product of two vectors A and B, written A • B, is the number
AB cos 6, where A and B are the magnitudes of the vectors A and B, and 6 is the angle
between them. Expressed in terms of Cartesian components of the two vectors,
A • B = A X B X + Ay By + A Z B Z .
1.5 Energy of a system of charges
13
ds = dr/ cos 0; hence F ■ ds = F dr. Each increment of work along one
path is matched by a corresponding increment on the other, so the sums
must be equal. Our conclusion holds even for paths that loop in and out,
like the dotted path in Fig. 1.5. (Why?)
Returning now to the two charges as we left them in Fig. 1.4(b), let
us bring in from some remote place a third charge qj, and move it to a
point F 3 whose distance from charge 1 is 7 - 31 , and from charge 2, r 32 - The
work required to effect this will be
rP 3
W 3 = - F 3 • ds. ( 1 . 10 )
J OO
Thanks to the additivity of electrical interactions, which we have already
emphasized,
— J F3 • ds — — J (F31 + F32) • ds
= - j F 31 ■ ds - j F 32 • ds. ( 1 . 11 )
That is, the work required to bring q 3 to P 3 is the sum of the work needed
when q\ is present alone and that needed when q 2 is present alone:
Figure 1.5.
Because the force is central, the sections of
different paths between r + dr and r require the
same amount of work.
W 3 =
1
1
qiq-i
4jreo r 3 i 4;reo r 32
( 1 . 12 )
The total work done in assembling this arrangement of three charges,
which we shall call U, is therefore
= 1 / qiqi qiqj
4?reo \ r l2 n 3
(1.13)
We note that q\, q 2 , and c /3 appear symmetrically in the expression
above, in spite of the fact that q 3 was brought in last. We would have
reached the same result if c /3 had been brought in first. (Try it.) Thus U is
independent of the order in which the charges were assembled. Since it
is independent also of the route by which each charge was brought in, U
must be a unique property of the final arrangement of charges. We may
call it the electrical potential energy of this particular system. There is
a certain arbitrariness, as always, in the definition of a potential energy.
In this case we have chosen the zero of potential energy to correspond to
the situation with the three charges already in existence but infinitely far
apart from one another. The potential energy belongs to the configuration
as a whole. There is no meaningful way of assigning a certain fraction
of it to one of the charges.
It is obvious how this very simple result can be generalized to apply
to any number of charges. If we have N different charges, in any arrange¬
ment in space, the potential energy of the system is calculated by sum¬
ming over all pairs, just as in Eq. (1.13). The zero of potential energy, as
in that case, corresponds to all charges far apart.
14
Electrostatics: charges and fields
(a)
— e
+le
b
— e
(b)
12 such pairs
Figure 1.6.
(a) The potential energy of this arrangement of
nine point charges is given by Eq. (1.14).
(b) Four types of pairs are involved in the sum.
Example (Charges in a cube) What is the potential energy of an arrange¬
ment of eight negative charges on the corners of a cube of side b, with a positive
charge in the center of the cube, as in Fig. 1.6(a)? Suppose each negative charge
is an electron with charge —e, while the central particle carries a double positive
charge, 2e.
Solution Figure 1.6(b) shows that there are four different types of pairs. One
type involves the center charge, while the other three involve the various edges
and diagonals of the cube. Summing over all pairs yields
1 / (—2e 2 ) e 2 e~ e
U = -- 8 ■ v + 12 ■ — + 12 • + 4 ■ —
47T6 0 \ (V3/2 )b b V2 b V3b
1 4.32e 2
4jreo b
(1.14)
The energy is positive, indicating that work had to be done on the system to
assemble it. That work could, of course, be recovered if we let the charges move
apart, exerting forces on some external body or bodies. Or if the electrons were
simply to fly apart from this configuration, the total kinetic energy of all the
particles would become equal to U. This would be true whether they came apart
simultaneously and symmetrically, or were released one at a time in any order.
Flere we see the power of this simple notion of the total potential energy of the
system. Think what the problem would be like if we had to compute the resultant
vector force on every particle at every stage of assembly of the configuration!
In this example, to be sure, the geometrical symmetry would simplify that task;
even so, it would be more complicated than the simple calculation above.
One way of writing the instruction for the sum over pairs is this:
1 N
Wk
2 •“ 4tt e 0 r jk
]= 1 kft J
(1.15)
The double-sum notation, Y-!j= i sa y s: ta ^ e ./ = 1 and sum over
k = 2, 3, 4,. .., N", then take j — 2 and sum over k = 1, 3, 4,..., N m , and so
on, through j = N. Clearly this includes every pair twice, and to correct
for that we put in front the factor 1 /2.
1.6 Electrical energy in a crystal lattice
These ideas have an important application in the physics of crystals. We
know that an ionic crystal like sodium chloride can be described, to a
very good approximation, as an arrangement of positive ions (Na + ) and
negative ions (Cl - ) alternating in a regular three-dimensional array or
lattice. In sodium chloride the arrangement is that shown in Fig. 1.7(a).
Of course the ions are not point charges, but they are nearly spherical
distributions of charge and therefore (as we shall prove in Section 1.11)
the electrical forces they exert on one another are the same as if each ion
1.6 Electrical energy in a crystal lattice
15
were replaced by an equivalent point charge at its center. We show this
electrically equivalent system in Fig. 1.7(b). The electrostatic potential
energy of the lattice of charges plays an important role in the explanation
of the stability and cohesion of the ionic crystal. Let us see if we can
estimate its magnitude.
We seem to be faced at once with a sum that is enormous, if not dou¬
bly infinite; any macroscopic crystal contains 10 20 atoms at least. Will
the sum converge? Now what we hope to find is the potential energy per
unit volume or mass of crystal. We confidently expect this to be inde¬
pendent of the size of the crystal, based on the general argument that
one end of a macroscopic crystal can have little influence on the other.
Two grams of sodium chloride ought to have twice the potential energy
of one gram, and the shape should not be important so long as the sur¬
face atoms are a small fraction of the total number of atoms. We would
be wrong in this expectation if the crystal were made out of ions of one
sign only. Then, 1 g of crystal would carry an enormous electric charge,
and putting two such crystals together to make a 2 g crystal would take
a fantastic amount of energy. (You might estimate how much!) The sit¬
uation is saved by the fact that the crystal structure is an alternation of
equal and opposite charges, so that any macroscopic bit of crystal is very
nearly neutral.
To evaluate the potential energy we first observe that every positive
ion is in a position equivalent to that of every other positive ion. Further¬
more, although it is perhaps not immediately obvious from Fig. 1.7, the
arrangement of positive ions around a negative ion is exactly the same as
the arrangement of negative ions around a positive ion, and so on. Hence
we may take one ion as a center, it matters not which kind, sum over its
interactions with all the others, and simply multiply by the total number
of ions of both kinds. This reduces the double sum in Eq. (1.15) to a sin¬
gle sum and a factor N; we must still apply the factor 1 /2 to compensate
for including each pair twice. That is, the energy of a sodium chloride
lattice composed of a total of N ions is
N
1 qiqk
4jre 0 r\k '
(1.16)
Figure 1.7.
A portion of a sodium chloride crystal, with the
ions Na + and Cl - shown in about the right
relative proportions (a), and replaced by
equivalent point charges (b).
Taking the positive ion at the center as in Fig. 1.7(b), our sum runs over
all its neighbors near and far. The leading terms start out as follows:
1 1 / 6e 2 12e 2 8e 2
U _ n - - I -1- — - — -b • • •
2 Ane 0 \ a V2a V3 a
The first term comes from the 6 nearest chlorine ions, at distance a , the
second from the 12 sodium ions on the cube edges, and so on. It is clear,
incidentally, that this series does not converge absolutely, if we were so
16
Electrostatics: charges and fields
foolish as to try to sum all the positive terms first, that sum would diverge.
To evaluate such a sum, we should arrange it so that as we proceed
outward, including ever more distant ions, we include them in groups
that represent nearly neutral shells of material. Then if the sum is bro¬
ken off, the more remote ions that have been neglected will be such an
even mixture of positive and negative charges that we can be confident
their contribution would have been small. This is a crude way to describe
what is actually a somewhat more delicate computational problem. The
numerical evaluation of such a series is easily accomplished with a com¬
puter. The answer in this example happens to be
—0.8738Afe 2
U= -
4iteoa
(1.18)
Here N, the number of ions, is twice the number of NaCl molecules.
The negative sign shows that work would have to be done to take
the crystal apart into ions. In other words, the electrical energy helps to
explain the cohesion of the crystal. If this were the whole story, however,
the crystal would collapse, for the potential energy of the charge distri¬
bution is obviously lowered by shrinking all the distances. We meet here
again the familiar dilemma of classical - that is, nonquantum - physics.
No system of stationary particles can be in stable equilibrium, according
to classical laws, under the action of electrical forces alone; we will give
a proof of this fact in Section 2.12. Does this make our analysis useless?
Not at all. Remarkably, and happily, in the quantum physics of crystals
the electrical potential energy can still be given meaning, and can be
computed very much in the way we have learned here.
1.7 The electric field
Suppose we have some arrangement of charges, < 71 , q 2 , ■ ■ ■, q.\ ! , fixed in
space, and we are interested not in the forces they exert on one another,
but only in their effect on some other charge qu that might be brought
into their vicinity. We know how to calculate the resultant force on this
charge, given its position which we may specify by the coordinates x, y,
Z. The force on the charge qo is
F =
. N
1 d0dj r 0j
r 2 0j
(1.19)
where rq/ is the vector from the / 1 h charge in the system to the point
(x,y, z). The force is proportional to qo, so if we divide out qo we obtain
a vector quantity that depends only on the structure of our original system
of charges, q 1 ,..., q^, and on the position of the point ( x,y,z ). We call
this vector function of x, y, z the electric field arising from the qi,... ,q^
1.7 The electric field
17
and use the symbol E for it. The charges q \,..., we call sources of
the field. We may take as the definition of the electric field E of a charge
distribution, at the point (x,y, z),
E (pc,y,z)
1
4jre 0
N
E
The force on some other charge q at (x, y, z) is then
F = qE
( 1 . 20 )
( 1 . 21 )
Figure 1.8 illustrates the vector addition of the field of a point charge
of 2 C to the field of a point charge of — 1 C, at a particular point in space.
In the SI system of units, electric field strength is expressed in newtons
per unit charge, that is, newtons/coulomb. In Gaussian units, with the esu
as the unit of charge and the dyne as the unit of force, the electric field
strength is expressed in dynes/esu.
After the introduction of the electric potential in Chapter 2, we shall
have another, and completely equivalent, way of expressing the unit of
electric field strength; namely, volts/meter in SI units and statvolts/
centimeter in Gaussian units.
So far we have nothing really new. The electric field is merely another
way of describing the system of charges; it does so by giving the force
per unit charge, in magnitude and direction, that an exploring charge qo
would experience at any point. We have to be a little careful with that
interpretation. Unless the source charges are really immovable, the intro¬
duction of some finite charge qo may cause the source charges to shift
their positions, so that the field itself, as defined by Eq. (1.20), is dif¬
ferent. That is why we assumed fixed charges to begin our discussion.
People sometimes define the field by requiring qo to be an “infinitesi¬
mal” test charge, letting E be the limit of F /qo as qo —> 0. Any flavor of
rigor this may impart is illusory. Remember that in the real world we have
never observed a charge smaller than e\ Actually, if we take Eq. (1.20) as
our definition of E, without reference to a test charge, no problem arises
and the sources need not be fixed. If the introduction of a new charge
causes a shift in the source charges, then it has indeed brought about a
change in the electric field, and if we want to predict the force on the new
charge, we must use the new electric field in computing it.
Perhaps you still want to ask, what is an electric field? Is it some¬
thing real, or is it merely a name for a factor in an equation that has to be
multiplied by something else to give the numerical value of the force we
measure in an experiment? Two observations may be useful here. First,
since it works, it doesn’t make any difference. That is not a frivolous
answer, but a serious one. Second, the fact that the electric field vector
Figure 1.8.
The field at a point is the vector sum of the fields
of each of the charges in the system.
18
Electrostatics: charges and fields
(a)
v /
♦
* v
4 V
• Charge +3
O Charge -1
Figure 1.9.
(a) Field of a charge q i = 3. (b) Field of a
charge qi = - 1. Both representations are
necessarily crude and only roughly quantitative.
at a point in space is all we need know to predict the force that will act
on any charge at that point is by no means trivial. It might have been
otherwise! If no experiments had ever been done, we could imagine that,
in two different situations in which unit charges experience equal force,
test charges of strength 2 units might experience unequal forces, depend¬
ing on the nature of the other charges in the system. If that were true, the
field description wouldn’t work. The electric field attaches to every point
in a system a local property, in this sense: if we know E in some small
neighborhood, we know, without further inquiry, what will happen to
any charges in that neighborhood. We do not need to ask what produced
the field.
To visualize an electric field, you need to associate a vector, that is, a
magnitude and direction, with every point in space. We shall use various
schemes in this book, none of them wholly satisfactory, to depict vector
fields.
It is hard to draw in two dimensions a picture of a vector function
in three-dimensional space. We can indicate the magnitude and direction
of E at various points by drawing little arrows near those points, mak¬
ing the arrows longer where E is larger. Using this scheme, we show in
Fig. 1.9(a) the field of an isolated point charge of 3 units and in Fig. 1.9(b)
the field of a point charge of — 1 unit. These pictures admittedly add noth¬
ing whatsoever to our understanding of the field of an isolated charge;
anyone can imagine a simple radial inverse-square field without the help
of a picture. We show them in order to combine (side by side) the two
fields in Fig. 1.10, which indicates in the same manner the field of two
such charges separated by a distance a. All that Fig. 1.10 can show is the
field in a plane containing the charges. To get a full three-dimensional
representation, one must imagine the figure rotated around the symmetry
axis. In Fig. 1.10 there is one point in space where E is zero. As an
exercise, you can quickly figure out where this point lies. Notice also
that toward the edge of the picture the field points more or less radially
outward all around. One can see that at a very large distance from the
charges the field will look very much like the field from a positive point
charge. This is to be expected because the separation of the charges can¬
not make very much difference for points far away, and a point charge
of 2 units is just what we would have left if we superimposed our two
sources at one spot.
Another way to depict a vector field is to draw field lines. These are
simply curves whose tangent, at any point, lies in the direction of the
field at that point. Such curves will be smooth and continuous except at
singularities such as point charges, or points like the one in the example
of Fig. 1.10 where the field is zero. A field line plot does not directly give
7
Such a representation is rather clumsy at best. It is hard to indicate the point in space to
which a particular vector applies, and the range of magnitudes of E is usually so large
that it is impracticable to make the lengths of the arrows proportional to E.
1.7 The electric field
19
• Charge +3
o Charge -1
\
Figure 1.10.
The field in the vicinity of two charges, q\ = +3,
q% = - 1, is the superposition of the fields in
Figs. 1.9(a) and (b).
Figure 1.11.
Some field lines in the electric field around two
charges, q\ = +3, q 2 = -1.
the magnitude of the field, although we shall see that, in a general way,
the field lines converge as we approach a region of strong field and spread
apart as we approach a region of weak field. In Fig. 1.11 are drawn some
field lines for the same arrangement of charges as in Fig. 1.10, a positive
charge of 3 units and a negative charge of 1 unit. Again, we are restricted
20
Electrostatics: charges and fields
by the nature of paper and ink to a two-dimensional section through a
three-dimensional bundle of curves.
1.8 Charge distributions
This is as good a place as any to generalize from point charges to contin¬
uous charge distributions. A volume distribution of charge is described
by a scalar charge-density function p, which is a function of position,
with the dimensions charge/volume. That is, p times a volume element
gives the amount of charge contained in that volume element. The same
symbol is often used for mass per unit volume, but in this book we shall
always give charge per unit volume first call on the symbol p. If we
write p as a function of the coordinates x, y, z, then p(x,y,z) dxdydz is
the charge contained in the little box, of volume dxdydz, located at the
point (x,y,z).
On an atomic scale, of course, the charge density varies enormously
from point to point; even so, it proves to be a useful concept in that
domain. However, we shall use it mainly when we are dealing with large-
scale systems, so large that a volume element dv — dx dy dz can be quite
small relative to the size of our system, although still large enough to
contain many atoms or elementary charges. As we have remarked before,
we face a similar problem in defining the ordinary mass density of a
substance.
If the source of the electric field is to be a continuous charge distri¬
bution rather than point charges, we merely replace the sum in Eq. (1.20)
with the appropriate integral. The integral gives the electric field at
(x,y, z), which is produced by charges at other points (x',y',zf).
E(x, y, z)
1 f p{x',y',z')rdx' dy’ dz'
4 tt€o J r 2
( 1 . 22 )
This is a volume integral. Holding (x, y, z) fixed, we let the variables of
integration, x!, /, and z!, range over all space containing charge, thus
summing up the contributions of all the bits of charge. The unit vector
r points from (xf,y',z r ) to (x,y, z) - unless we want to put a minus sign
before the integral, in which case we may reverse the direction of r. It is
always hard to keep signs straight. Let’s remember that the electric field
points away from a positive source (Fig. 1.12).
Figure 1.12.
Each element of the charge distribution
p{x! ,/ ,z!) makes a contribution to the electric
field E at the point ( x,y,z ). The total field at this
point is the sum of all such contributions; see
Eq. (1.22).
Example (Field due to a hemisphere) A solid hemisphere has radius R
and uniform charge density p. Find the electric field at the center.
Solution Our strategy will be to slice the hemisphere into rings around the
symmetry axis. We will find the electric field due to each ring, and then integrate
over the rings to obtain the field due to the entire hemisphere. We will work with
1.8 Charge distributions
21
polar coordinates (or, equivalently, spherical coordinates), which are much more
suitable than Cartesian coordinates in this setup.
The cross section of a ring is (essentially) a little rectangle with side lengths
dr and rd9, as shown in Fig. 1.13. The cross-sectional area is thus rdrdd. The
radius of the ring is r sin#, so the volume is (rdrd9)(2nr sin 0). The charge in
the ring is therefore p(2nr 2 sin 9 dr dd). Equivalently, we can obtain this result
by using the standard spherical-coordinate volume element, r 2 sinO drdd dtj>,
and then integrating over (p to obtain the factor of 2n.
Consider a tiny piece of the ring, with charge dq. This piece creates an elec¬
tric field at the center of the hemisphere that points diagonally upward (if p is
positive) with magnitude dq/AneQi' 2 . However, only the vertical component sur¬
vives, because the horizontal component cancels with the horizontal component
from the diametrically opposite charge dq on the ring. The vertical component
involves a factor of cos 9. When we integrate over the whole ring, the dq simply
integrates to the total charge we found above. The (vertical) electric field due to
a given ring is therefore
Figure 1.13.
Cross section of a thin ring. The hemisphere
may be considered to be built up from rings.
p {In r~ sin 8 dr dS ) p sin 8 cos 9 dr dd
dE y = -r-cos 9 = -.
47r 6(p' 2 2eo
(1.23)
Integrating over r and 8 to obtain the field due to the entire hemisphere gives
r R f 71 / 2 p sin# cos0 drd9
y JO Jo 2e 0
sin 8 cos 9 dd
p sin 2 9 71/2 _ pR
2fq 2 o 4eo
(1.24)
Note that the radius r canceled in Eq. (1.23). For given values of 9, dd, and dr, the
volume of a ring grows like r 2 , and this exactly cancels the r 2 in the denominator
in Coulomb’s law.
Remark As explained above, the electric field due to the hemisphere is verti¬
cal. This fact also follows from considerations of symmetry. We will make many
symmetry arguments throughout this book, so let us be explicit here about how
the reasoning proceeds. Assume (in search of a contradiction) that the electric
field due to the hemisphere is not vertical. It must then point off at some angle,
as shown in Fig. 1.14(a). Let’s say that the E vector lies above a given dashed line
painted on the hemisphere. If we rotate the system by, say. 180° around the sym¬
metry axis, the field now points in the direction shown in Fig. 1.14(b), because
it must still pass over the dashed line. But we have exactly the same hemisphere
after the rotation, so the field must still point upward to the right. We conclude
that the field due to the hemisphere points both upward to the left and upward to
the right. This is a contradiction. The only way to avoid this contradiction is for
the field to point along the symmetry axis (possibly in the negative direction),
because in that case it doesn’t change under the rotation.
In the neighborhood of a true point charge the electric field grows
infinite like 1 /r 2 as we approach the point. It makes no sense to talk about
the field at the point charge. As our ultimate physical sources of field are
(b)
Figure 1.14.
The symmetry argument that explains why E
must be vertical.
22
Electrostatics: charges and fields
(b)
(c)
Figure 1.15.
(a) A closed surface in a vector field is divided
(b) into small elements of area, (c) Each
element of area is represented by an outward
vector.
not, we believe, infinite concentrations of charge in zero volume, but
instead finite structures, we simply ignore the mathematical singularities
implied by our point-charge language and rule out of bounds the interior
of our elementary sources. A continuous charge distribution p{x !,/,z')
that is nowhere infinite gives no trouble at all. Equation (1.22) can be
used to find the field at any point within the distribution. The integrand
doesn’t blow up at r = 0 because the volume element in the numerator
equals r 2 sin cp d(f> d6 dr in spherical coordinates, and the r 2 here can¬
cels the r 2 in the denominator in Eq. (1.22). That is to say, so long as p
remains finite, the field will remain finite everywhere, even in the interior
or on the boundary of a charge distribution.
1.9 Flux
The relation between the electric field and its sources can be expressed
in a remarkably simple way, one that we shall find very useful. For this
we need to define a quantity called flux.
Consider some electric field in space and in this space some arbi¬
trary closed surface, like a balloon of any shape. Figure 1.15 shows such
a surface, the field being suggested by a few field lines. Now divide the
whole surface into little patches that are so small that over any one patch
the surface is practically flat and the vector field does not change appre¬
ciably from one part of a patch to another. In other words, don’t let the
balloon be too crinkly, and don’t let its surface pass right through a sin¬
gularity 8 of the field such as a point charge. The area of a patch has a
certain magnitude in square meters, and a patch defines a unique direc¬
tion - the outward-pointing normal to its surface. (Since the surface is
closed, you can tell its inside from its outside; there is no ambiguity.) Let
this magnitude and direction be represented by a vector. Then for every
patch into which the surface has been divided, such as patch number j,
we have a vector ay giving its area and orientation. The steps we have just
taken are pictured in Figs. 1.15(b) and (c). Note that the vector ay does
not depend at all on the shape of the patch; it doesn’t matter how we have
divided up the surface, as long as the patches are small enough.
Let Ey be the electric field vector at the location of patch number
j. The scalar product Ey ■ a, is a number. We call this number the flux
through that bit of surface. To understand the origin of the name, imagine
a vector function that represents the velocity of motion in a fluid - say in a
river, where the velocity varies from one place to another but is constant
in time at any one position. Denote this vector field by v, measured in
8 By a singularity of the field we would ordinarily mean not only a point source where
the field approaches infinity, but also any place where the field changes magnitude or
direction discontinuously, such as an infinitesimally thin layer of concentrated charge.
Actually this latter, milder, kind of singularity would cause no difficulty here unless
our balloon’s surface were to coincide with the surface of discontinuity over some
finite area.
1.10 Gauss’s law
23
Flux = va
Flux = 0
Flux = va cos 60° = 0.5va
meters/second. Then, if a is the oriented area in square meters of a frame
lowered into the water, v ■ a is the rate of flow of water through the frame
in cubic meters per second (Fig. 1.16). The cos 0 factor in the standard
expression for the dot product correctly picks out the component of v
along the direction of a, or equivalently the component of a along the
direction of v. We must emphasize that our definition of flux is applicable
to any vector function, whatever physical variable it may represent.
Now let us add up the flux through all the patches to get the flux
through the entire surface, a scalar quantity which we shall denote by <t>:
Figure 1.16.
The flux through the frame of area a is v • a,
where v is the velocity of the fluid. The flux is the
volume of fluid passing through the frame, per
unit time.
° = E E /" a /-
ally
(1.25)
Letting the patches become smaller and more numerous without limit,
we pass from the sum in Eq. (1.25) to a surface integral:
0= / E • c/a. (1.26)
J entire
surface
A surface integral of any vector function F, over a surface S, means just
this: divide S into small patches, each represented by a vector outward, of
magnitude equal to the patch area; at every patch, take the scalar product
of the patch area vector and the local F; sum all these products, and the
limit of this sum, as the patches shrink, is the surface integral. Do not
be alarmed by the prospect of having to perform such a calculation for
an awkwardly shaped surface like the one in Fig. 1.15. The surprising
property we are about to demonstrate makes that unnecessary!
1.10 Gauss’s law
Take the simplest case imaginable; suppose the field is that of a single
isolated positive point charge q, and the surface is a sphere of radius r
centered on the point charge (Fig. 1.17). What is the flux <J> through this
surface? The answer is easy because the magnitude of E at every point
on the surface is q/ATte^r 2 and its direction is the same as that of the
outward normal at that point. So we have
T> = E ■ (total area) = —-—■ 4 nr 2 = —. (1-27)
47reo H eo
Figure 1.17.
In the field E of a point charge q , what is the
outward flux over a sphere surrounding q?
24
Electrostatics: charges and fields
The flux is independent of the size of the sphere. Here for the first time
we see the benefit of including the factor of 1/4jr in Coulomb’s law
in Eq. (1.4). Without this factor, we would have an uncanceled factor
of 4 it in Eq. (1.27) and therefore also, eventually, in one of Maxwell’s
equations. Indeed, in Gaussian units Eq. (1.27) takes the form of
cp = 4nq.
Now imagine a second surface, or balloon, enclosing the first, but
not spherical, as in Fig. 1.18. We claim that the total flux through this
surface is the same as that through the sphere. To see this, look at a cone,
radiating from q, that cuts a small patch a out of the sphere and continues
on to the outer surface, where it cuts out a patch A at a distance R from
the point charge. The area of the patch A is larger than that of the patch
a by two factors: first, by the ratio of the distance squared ( R/r) 2 \ and
second, owing to its inclination, by the factor 1 / cos 9. The angle 9 is the
angle between the outward normal and the radial direction (see Fig. 1.18).
The electric field in that neighborhood is reduced from its magnitude on
the sphere by the factor ( r/R) 2 and is still radially directed. Letting Eq?)
be the field at the outer patch and Eq) be the field at the sphere, we have
flux through outer patch = Eq?) • A = E(^A cos 9,
flux through inner patch = E( r ) ■ a = E( r) a. (1-28)
Using the above facts concerning the magnitude of Eq?) and the area of
A, the flux through the outer patch can be written as
Figure 1.18.
Showing that the flux through any closed
surface around q is the same as the flux through
the sphere.
Eqt)A cos 9
cos 9 — E( r )a,
(1.29)
which equals the flux through the inner patch.
Now every patch on the outer surface can in this way be put into
correspondence with part of the spherical surface, so the total flux must
be the same through the two surfaces. That is, the flux through the new
surface must be just q/e o- But this was a surface of arbitrary shape and
size. We conclude: the flux of the electric field through any surface
enclosing a point charge q is q/e o. As a corollary we can say that the
total flux through a closed surface is zero if the charge lies outside the
surface. We leave the proof of this to the reader, along with Fig. 1.19 as
a hint of one possible line of argument.
There is a way of looking at all this that makes the result seem obvi¬
ous. Imagine at q a source that emits particles - such as bullets or photons
- in all directions at a steady rate. Clearly the flux of particles through a
window of unit area will fall off with the inverse square of the window’s
distance from q. Hence we can draw an analogy between the electric field
strength E and the intensity of particle flow in bullets per unit area per
9
To be sure, we had the second surface enclosing the sphere, but it didn’t have to, really.
Besides, the sphere can be taken as small as we please.
1.10 Gauss’s law
25
unit time. It is pretty obvious that the flux of bullets through any surface
completely surrounding q is independent of the size and shape of that
surface, for it is just the total number emitted per unit time. Correspond¬
ingly, the flux of E through the closed surface must be independent of
size and shape. The common feature responsible for this is the inverse-
square behavior of the intensity.
The situation is now ripe for superposition! Any electric field is the
sum of the fields of its individual sources. This property was expressed
in our statement, Eq. (1.19), of Coulomb’s law. Clearly flux is an addi¬
tive quantity in the same sense, for if we have a number of sources,
q\,q 2 ,..., qN, the fields of which, if each were present alone, would be
Ej, E 2 ,..., EjV, then the flux <t> through some surface S in the actual field
can be written
$ = / E ■ da = I (Ei + E 2 + • • • + Ejv) • da. (1.30)
Js Js
We have just learned that f s E,- • da equals qi/e 0 if the charge qj
is inside S and equals zero otherwise. So every charge q inside the sur¬
face contributes exactly q/e 0 to the surface integral of Eq. (1.30) and all
charges outside contribute nothing. We have arrived at Gauss’s law.
The flux of the electric field E through any closed surface, that is,
the integral fE-da over the surface, equals 1 /eo times the total
charge enclosed by the surface:
/ E • da = — Y qt = — [ pdv (Gauss’s law) (1.31)
eo e 0 J
l
(a)
Figure 1.19.
To show that the flux through the closed surface
in (a) is zero, you can make use of (b).
We call the statement in the box a law because it is equivalent to
Coulomb’s law and it could serve equally well as the basic law of elec¬
trostatic interactions, after charge and field have been defined. Gauss’s
law and Coulomb’s law are not two independent physical laws, but the
same law expressed in different ways. 1 " In Gaussian units, the 1/eo in
Gauss’s law is replaced with 47r.
Looking back over our proof, we see that it hinged on the inverse-
square nature of the interaction and of course on the additivity of
interactions, or superposition. Thus the theorem is applicable to any
inverse-square field in physics, for instance to the gravitational field.
10 There is one difference, inconsequential here, but relevant to our later study of the
fields of moving charges. Gauss’s law is obeyed by a wider class of fields than those
represented by the electrostatic field. In particular, a field that is inverse-square in r but
not spherically symmetrical can satisfy Gauss’s law. In other words, Gauss’s law alone
does not imply the symmetry of the field of a point source which is implicit in
Coulomb’s law.
26
Electrostatics: charges and fields
It is easy to see that Gauss’s law would not hold if the law of force
were, say, inverse-cube. For in that case the flux of electric field from
a point charge q through a sphere of radius R centered on the charge
would be
$ = / E da = —-— 5 - • 4ttR 2
J 4jte 0 R 3
q
eo R
(1.32)
Figure 1.20.
A charge distribution with spherical symmetry.
Figure 1.21.
The electric field of a spherical charge
distribution.
By making the sphere large enough we could make the flux through it as
small as we pleased, while the total charge inside remained constant.
This remarkable theorem extends our knowledge in two ways. First,
it reveals a connection between the field and its sources that is the con¬
verse of Coulomb’s law. Coulomb’s law tells us how to derive the elec¬
tric field if the charges are given; with Gauss’s law we can determine how
much charge is in any region if the field is known. Second, the mathemat¬
ical relation here demonstrated is a powerful analytic tool; it can make
complicated problems easy, as we shall see in the following examples. In
Sections 1.11-1.13 we use Gauss’s law to calculate the electric field due
to various nicely shaped objects. In all of these examples the symmetry
of the object will play a critical role.
1.11 Field of a spherical charge distribution
We can use Gauss’s law to find the electric field of a spherically sym¬
metrical distribution of charge, that is, a distribution in which the charge
density p depends only on the radius from a central point. Figure 1.20
depicts a cross section through some such distribution. Here the charge
density is high at the center, and is zero beyond ro- What is the electric
field at some point such as Pi outside the distribution, or P 2 inside it
(Fig. 1.21)? If we could proceed only from Coulomb’s law, we should
have to carry out an integration that would sum the electric field vectors
at P[ arising from each elementary volume in the charge distribution.
Let’s try a different approach that exploits both the symmetry of the sys¬
tem and Gauss’s law.
Because of the spherical symmetry, the electric field at any point
must be radially directed - no other direction is unique. Likewise, the
field magnitude E must be the same at all points on a spherical surface S\
of radius r \, for all such points are equivalent. Call this field magnitude
£ 1 . The flux through this surface .S) is therefore simply 4itr\E\. and by
Gauss’s law this must be equal to 1 /eo times the charge enclosed by the
surface. That is, 4nr^E\ = (1/eo) • (charge inside 5)) or
charge inside S \
E 1 = ---^-. (1.33)
47 r e 0 rj-
Comparing this with the field of a point charge, we see that the field
at all points on Si is the same as if all the charge within Si were con¬
centrated at the center. The same statement applies to a sphere drawn
1.11 Field of a spherical charge distribution
27
inside the charge distribution. The field at any point on S 2 is the same as
if all charge within S 2 were at the center, and all charge outside S 2 absent.
Evidently the field inside a “hollow” spherical charge distribution is zero
(Fig. 1.22). Problem 1.17 gives an alternative derivation of this fact.
Example (Field inside and outside a uniform sphere) A spherical
charge distribution has a density p that is constant from r= 0 out to r=R
and is zero beyond. What is the electric field for all values of r, both less than
and greater than R1
Solution For r > R, the field is the same as if all of the charge were concen¬
trated at the center of the sphere. Since the volume of the sphere is 4nR 3 /3, the
field is therefore radial and has magnitude
E(r) =
(4nR 3 /3)p
4neor 2
pR 3
3 e 0 r 2
{r > R).
(1.34)
For r < R, the charge outside radius r effectively contributes nothing to the field, Figure 1.22.
while the charge inside radius r acts as if it were concentrated at the center. The The field is zero inside a spherical shell of
volume inside radius r is 4nr 3 /3, so the field inside the given sphere is radial charge,
and has magnitude
E(r) =
{4nr 3 / 3) p
4n eyr 2
Pr
3«o
= f- (r< R).
(1.35)
In terms of the total charge Q = (4jiR 3 /3)p, this can be written as Qr/47ieoR?.
The field increases linearly with r inside the sphere; the r 3 growth of the effec¬
tive charge outweighs the 1 /r 2 effect from the increasing distance. And the field
decreases like 1 / r 2 outside the sphere. A plot of E(r) is shown in Fig. 1.23. Note
that E(r) is continuous at r = R, where it takes on the value pR/3(Q. As we will
see in Section 1.13, field discontinuities are created by surface charge densities,
and there are no surface charges in this system. The field goes to zero at the cen¬
ter, so it is continuous there also. How should the density vary with r so that the
magnitude E(r) is uniform inside the sphere? That is the subject of Exercise 1.68.
Figure 1.23.
The electric field due to a uniform sphere of
charge.
The same argument applied to the gravitational field would tell us
that the earth, assuming it is spherically symmetrical in its mass distribu¬
tion, attracts outside bodies as if its mass were concentrated at the center.
That is a rather familiar statement. Anyone who is inclined to think the
principle expresses an obvious property of the center of mass must be
reminded that the theorem is not even true, in general, for other shapes.
A perfect cube of uniform density does not attract external bodies as if
its mass were concentrated at its geometrical center.
Newton didn’t consider the theorem obvious. He needed it as the
keystone of his demonstration that the moon in its orbit around the earth
and a falling body on the earth are responding to similar forces. The delay
of nearly 20 years in the publication of Newton’s theory of gravitation
28
Electrostatics: charges and fields
Figure 1.24.
(a) The field at P is the vector sum of
contributions from each element of the line
charge, (b) Detail of (a).
was apparently due, in part at least, to the trouble he had in proving this
theorem to his satisfaction. The proof he eventually devised and pub¬
lished in the Principia in 1686 (Book I, Section XII, Theorem XXXI)
is a marvel of ingenuity in which, roughly speaking, a tricky volume
integration is effected without the aid of the integral calculus as we
know it. The proof is a good bit longer than our whole preceding dis¬
cussion of Gauss’s law, and more intricately reasoned. You see, with all
his mathematical resourcefulness and originality, Newton lacked Gauss’s
law - a relation that, once it has been shown to us, seems so obvious as
to be almost trivial.
1.12 Field of a line charge
A long, straight, charged wire, if we neglect its thickness, can be charac¬
terized by the amount of charge it carries per unit length. Let k, measured
in coulombs/meter, denote this linear charge density. What is the elec¬
tric field of such a line charge, assumed infinitely long and with constant
linear charge density kl We’ll do the problem in two ways, first by an
integration starting from Coulomb’s law, and then by using Gauss’s law.
To evaluate the field at the point P, shown in Fig. 1.24, we must add
up the contributions from all segments of the line charge, one of which
is indicated as a segment of length dx. The charge dq on this element is
given by dq — k dx. Having oriented our x axis along the line charge, we
may as well let the y axis pass through P, which is a distance r from the
nearest point on the line. It is a good idea to take advantage of symmetry
at the outset. Obviously the electric field at P must point in the y direc¬
tion, so that E x and E z are both zero. The contribution of the charge dq
to the y component of the electric field at P is
dEy
dq
AtzcqR 2
cos 9 =
k dx
- - „ cos <9,
47T€oR 2
(1.36)
where 9 is the angle the electric field of dq makes with the y direction.
The total y component is then
Ey =
k cos 9
AttcqR 2
dx.
(1.37)
It is convenient to use 6 as the variable of integration. Since Figs. 1.24(a)
and (b) tell us that R = r j cos 0 and dx = R dO / cos 0, we have dx =
r dO/ cos 2 9. (This expression for dx comes up often. It also follows from
x = rtan0 => dx = rrf(tand) = r d9/ cos 2 9.) Eliminating dx and R
from the integral in Eq. (1.37), in favor of 9, we obtain
Ey =
/-
71 ! 2 k cos 9 d9
—tt /2 4jreo r
k ri 2 k
- / cos 9 d9 = -.
4ite 0 r j— 77-/2 27T€ 0 r
(1.38)
We see that the field of an infinitely long, uniformly dense line charge is
proportional to the reciprocal of the distance from the line. Its direction
1.13 Field of an infinite flat sheet of charge
29
is of course radially outward if the line carries a positive charge, inward
if negative.
Gauss’s law leads directly to the same result. Surround a segment of
the line charge with a closed circular cylinder of length L and radius r,
as in Fig. 1.25, and consider the flux through this surface. As we have
already noted, symmetry guarantees that the field is radial, so the flux
through the ends of the “tin can” is zero. The flux through the cylindrical
surface is simply the area, IjxrL, times E r , the field at the surface. On the
other hand, the charge enclosed by the surface is just XL, so Gauss’s law
gives us ( 2jtrL)E r = XL/e o or
E r = (1.39)
lne^r
in agreement with Eq. (1.38).
Figure 1.25.
Using Gauss’s law to find the field of a line
charge.
1.13 Field of an infinite flat sheet of charge
Electric charge distributed smoothly in a thin sheet is called a surface
charge distribution. Consider a flat sheet, infinite in extent, with the con¬
stant surface charge density a. The electric field on either side of the
sheet, whatever its magnitude may turn out to be, must surely point per¬
pendicular to the plane of the sheet; there is no other unique direction
in the system. Also, because of symmetry, the field must have the same
magnitude and the opposite direction at two points P and P' equidistant
from the sheet on opposite sides. With these facts established. Gauss’s
law gives us at once the field intensity, as follows: draw a cylinder, as in
Fig. 1.26 (actually, any shape with uniform cross section will work fine),
with P on one side and P' on the other, of cross-sectional area A. The
outward flux is found only at the ends, so that if Ep denotes the magni¬
tude of the field at P, and Ep> the magnitude at P', the outward flux is
AEp + A Ep: = 2 AEp. The charge enclosed is a A, so Gauss’s law gives
2 AEp — oA/e o, or
Ep = (1.40)
2e 0
We see that the field strength is independent of r, the distance from the
sheet. Equation (1.40) could have been derived more laboriously by cal¬
culating the vector sum of the contributions to the field at P from all the
little elements of charge in the sheet.
In the more general case where there are other charges in the vicinity,
the field need not be perpendicular to the sheet, or symmetric on either
side of it. Consider a very squat Gaussian surface, with P and P' infinites¬
imally close to the sheet, instead of the elongated surface in Fig. 1.26.
We can then ignore the negligible flux through the cylindrical “side” of
the pillbox, so the above reasoning gives E±j> + Ej_ p/ = cr/eo, where
the “_L” denotes the component perpendicular to the sheet. If you want
CT(C/m 2 )
Figure 1.26.
Using Gauss’s law to find the field of an infinite
flat sheet of charge.
30
Electrostatics: charges and fields
to write this in terms of vectors, it becomes Ej_,p — Ej_ p/ = (er/eo)n,
where n is the unit vector perpendicular to the sheet, in the direction of
P. In other words, the discontinuity in Ej_ across the sheet is given by
AEj_ = —n. (1.41)
eo
Only the normal component is discontinuous; the parallel component is
continuous across the sheet. So we can just as well replace the AEj_ in
Eq. (1.41) with AE. This result is also valid for any finite-sized sheet,
because from up close the sheet looks essentially like an infinite plane,
at least as far as the normal component is concerned.
The field of an infinitely long line charge, we found, varies inversely
as the distance from the line, while the field of an infinite sheet has the
same strength at all distances. These are simple consequences of the fact
that the field of a point charge varies as the inverse square of the distance.
If that doesn’t yet seem compellingly obvious, look at it this way: roughly
speaking, the part of the line charge that is mainly responsible for the
field at P in Fig. 1.24 is the near part - the charge within a distance of
order of magnitude r. If we lump all this together and forget the rest, we
have a concentrated charge of magnitude q kr, which ought to produce
a field proportional to q/r 2 , or k/r. In the case of the sheet, the amount
of charge that is “effective,” in this sense, increases proportionally to r 2
as we go out from the sheet, which just offsets the 1 / r 2 decrease in the
field from any given element of charge.
m *
ft
Figure 1.27.
A spherical surface with uniform charge
density a.
1.14 The force on a layer of charge
The sphere in Fig. 1.27 has a charge distributed over its surface with
the uniform density er, in C/m 2 . Inside the sphere, as we have already
learned, the electric field of such a charge distribution is zero. Outside
the sphere the field is Q/Ane^r 2 , where Q is the total charge on the
sphere, equal to Artr^a. So just outside the surface of the sphere the field
strength is
a
Fjust outside — ■ (1-42)
eo
Compare this with Eq. (1.40) and Fig. 1.26. In both cases Gauss’s law is
obeyed: the change in the normal component of E, from one side of the
layer to the other, is equal to er/eo, in accordance with Eq. (1.41).
What is the electrical force experienced by the charges that make up
this distribution? The question may seem puzzling at first because the
field E arises from these very charges. What we must think about is the
force on some small element of charge dq, such as a small patch of area
dA with charge dq — odA. Consider, separately, the force on dq due to all
the other charges in the distribution, and the force on the patch due to the
charges within the patch itself. This latter force is surely zero. Coulomb
repulsion between charges within the patch is just another example of
1.14 The force on a layer of charge
31
Newton’s third law; the patch as a whole cannot push on itself. That
simplifies our problem, for it allows us to use the entire electric field E,
including the field due to all charges in the patch, in calculating the force
dF on the patch of charge dq:
dF = Edq = FodA. (1.43)
But what E shall we use, the field E = a /eo outside the sphere or the
field E = 0 inside? The correct answer, as we shall prove in a moment,
is the average of the two fields that is.
dF — - (er/eo + 0)cr dA
a 2 dA
2eo
(1.44)
(a)
A r
£•=()-
■ E = a/e 0
pAr = a
To justify this we shall consider a more general case, and one that
will introduce a more realistic picture of a layer of surface charge. Real (b)
charge layers do not have zero thickness. Figure 1.28 shows some ways
in which charge might be distributed through the thickness of a layer. In
each example, the value of a, the total charge per unit area of layer, is
the same. These might be cross sections through a small portion of the
spherical surface in Fig. 1.27 on a scale such that the curvature is not
noticeable. To make it more general, however, we can let the field on the
left be E\ (rather than 0, as it was inside the sphere), with £3 the held on
the right. The condition imposed by Gauss’s law, for given a, is, in each
case.
A r
-E = o/e 0
E2 — Ei — —.
eo
(1.45)
E = 0-
Now let us look carefully within the layer where the field is changing
continuously from E\ to £2 and there is a volume charge density p(x) (c)
extending from x = 0 to x = xq, the thickness of the layer (Fig. 1.29).
Consider a much thinner slab, of thickness dx dd xq, which contains per
unit area an amount of charge p dx. If the area of this thin slab is A, the
force on it is
dF = Ep dx ■ A.
(1.46)
Thus the total force per unit area of our original charge layer is
£
A
St-L
— = / Ep dx.
(1.47)
But Gauss’s law tells us via Eq. (1.45) that dE, the change in £ through
the thin slab, is just p dx/eo. Hence p dx in Eq. (1.47) can be replaced by
60 dE, and the integral becomes
- A = f^oEdE=^(E 2 2-Ef).
E = 0
E = <J/e 0
Figure 1.28.
The net change in field at a charge layer
depends only on the total charge per unit area.
(1.48)
32
Electrostatics: charges and fields
Figure 1.29.
Within the charge layer of density p(x),
E{x + dx) — E{x) = p dx/e q.
Since E 2 — E\ = er/eo, the force per unit area in Eq. (1.48), after being
factored, can be expressed as
F
A
-(£1 +E 2 )c
(1.49)
We have shown, as promised, that for given a the force per unit area on
a charge layer is determined by the mean of the external field on one
side and that on the other. This is independent of the thickness of the
layer, as long as it is small compared with the total area, and of the vari¬
ation p(x ) in charge density within the layer. See Problem 1.30 for an
alternative derivation of Eq. (1.49).
The direction of the electrical force on an element of the charge on
the sphere is, of course, outward whether the surface charge is positive or
negative. If the charges do not fly off the sphere, that outward force must
be balanced by some inward force, not included in our equations, that
can hold the charge carriers in place. To call such a force “nonelectrical”
would be misleading, for electrical attractions and repulsions are the
dominant forces in the structure of atoms and in the cohesion of matter
generally. The difference is that these forces are effective only at short
distances, from atom to atom, or from electron to electron. Physics on
that scale is a story of individual particles. Think of a charged rubber
balloon, say 0.1 m in radius, with 10 _8 C of negative charge spread as
uniformly as possible on its outer surface. It forms a surface charge of
density a = (10 -8 C)/47r(0.1 m) 2 = 8 • 10 -8 C/m 2 . The resulting out¬
ward force, per area of surface charge, is given by Eq. (1.44) as
dF _ a 2
dA 2eo
(8 ■ 10 -8 C/m 2 ) 2
2(8.85- 10- 12 C 2 /(Nm 2 ))
3.6 • 10 _4 N/m 2 . (1.50)
In fact, our charge consists of about 6 • 10 10 electrons attached to the
rubber film, which corresponds to about 50 million extra electrons per
square centimeter. So the “graininess” in the charge distribution is hardly
apparent. However, if we could look at one of these extra electrons, we
would find it roughly 10 _4 cm - an enormous distance on an atomic
scale - from its nearest neighbor. This electron would be stuck, elec¬
trically stuck, to a local molecule of rubber. The rubber molecule would
be attached to adjacent rubber molecules, and so on. If you pull on the
electron, the force is transmitted in this way to the whole piece of rubber.
Unless, of course, you pull hard enough to tear the electron loose from
the molecule to which it is attached. That would take an electric field
many thousands of times stronger than the field in our example.
Note that this is not necessarily the same as the average field within the layer, a
quantity of no special interest or significance.
1.15 Energy associated with the electric field
33
1.15 Energy associated with the electric field
Suppose our spherical shell of charge is compressed slightly, from an
initial radius of ro to a smaller radius, as in Fig. 1.30. This requires
that work be done against the repulsive force, which we found above to
be a 2 /2eo newtons for each square meter of surface. The displacement
being dr, the total work done is (47rr^)(<r 2 /2eo) dr, or (2jr r^a 2 /eo) dr.
This represents an increase in the energy required to assemble the system
of charges, the energy U we talked about in Section 1.5:
2jt rj.a 2
dU= - ?—dr. (1.51)
eo
Notice how the electric field E has been changed. Within the shell of
thickness dr, the field was zero and is now a/e o- Beyond ro the field is
unchanged. In effect we have created a field of strength E = a/e o filling
a region of volume Arc r^ dr. We have done so by investing an amount
of energy given by Eq. (1.51) which, if we substitute e^E for a, can be
written like this:
Figure 1.30.
Shrinking a spherical shell or charged balloon.
dU =
eo E 2
2
AnrQ dr.
(1.52)
This is an instance of a general theorem which we shall not prove
now (but see Problem 1.33): the potential energy U of a system of charges,
which is the total work required to assemble the system, can be calculated
from the electric field itself simply by assigning an amount of energy
(eoE 2 /2) dv to every volume element dv and integrating over all space
where there is electric field'.
U=^ E 2 dv
2 /entire
field
(1.53)
E 2 is a scalar quantity, of course: E 2 = E • E.
One may think of this energy as “stored” in the field. The system
being conservative, that amount of energy can of course be recovered
by allowing the charges to go apart; so it is nice to think of the energy
as “being somewhere” meanwhile. Our accounting comes out right if
we think of it as stored in space with a density of cqE 2 /2, in joules/m 3 .
There is no harm in this, but in fact we have no way of identifying, quite
independently of anything else, the energy stored in a particular cubic
meter of space. Only the total energy is physically measurable, that is,
the work required to bring the charge into some configuration, starting
from some other configuration. Just as the concept of electric field serves
in place of Coulomb’s law to explain the behavior of electric charges, so
when we use Eq. (1.53) rather than Eq. (1.15) to express the total poten¬
tial energy of an electrostatic system, we are merely using a different
kind of bookkeeping. Sometimes a change in viewpoint, even if it is at
34
Electrostatics: charges and fields
first only a change in bookkeeping, can stimulate new ideas and deeper
understanding. The notion of the electric field as an independent entity
will take form when we study the dynamical behavior of charged matter
and electromagnetic radiation.
Example (Potential energy of a uniform sphere) What is the energy
stored in a sphere of radius R with charge Q uniformly distributed throughout
the interior?
Solution The electric field is nonzero both inside and outside the sphere, so
Eq. (1.53) involves two different integrals. Outside the sphere, the field at radius
r is simply Q/4neQi' 2 , so the energy stored in the external field is
(1.54)
The example in Section 1.11 gives the field at radius r inside the sphere as
E r = pr/ 3eo- But the density equals p = Q/(4tzR 2 /3), so the field is E r =
{3Q/4nR?)r/3eQ = Qr/Ane^R?. The energy stored in the internal field is
therefore
(1.55)
This is one-fifth of the energy stored in the external field. The total energy is
the sum of (7 ex t an d Cj nt , which we can write as (3/5)Q 2 /47T€qR. We see that it
takes three-fifths as much energy to assemble the sphere as it does to bring in two
point charges Q to a separation of R. Exercise 1.61 presents an alternative method
of calculating the potential energy of a uniformly charged sphere, by imagining
building it up layer by layer.
We run into trouble if we try to apply Eq. (1.53) to a system that
contains a point charge, that is, a finite charge q of zero size. Locate q
at the origin of the coordinates. Close to the origin, E 2 will approach
q 2 /(4jT€o) 2 r 4 . With dv = Ajtr 2 dr, the integrand E 2 dv will behave like
dr/r 2 , and our integral will blow up at the limit r = 0. That simply tells
us that it would take infinite energy to pack finite charge into zero volume
- which is true but not helpful. In the real world we deal with particles
like electrons and protons. They are so small that for most purposes we
can ignore their dimensions and think of them as point charges when we
consider their electrical interaction with one another. How much energy
it took to make such a particle is a question that goes beyond the range of
classical electromagnetism. We have to regard the particles as supplied
to us ready-made. The energy we are concerned with is the work done in
moving them around.
The distinction is usually clear. Consider two charged particles, a
proton and a negative pion, for instance. Let E p be the electric field of
the proton, E T that of the pion. The total field is L = K p + E T , and E ■ E
1.16 Applications
35
equals + 2E p • E w . According to Eq. (1.53) the total energy in
the electric field of this two-particle system is
(1.56)
The value of the first integral is a property of any isolated proton. It is
a constant of nature which is not changed by moving the proton around.
The same goes for the second integral, involving the pion’s electric field
alone. It is the third integral that directly concerns us, for it expresses
the energy required to assemble the system given a proton and a pion as
constituents.
The distinction could break down if the two particles interact so
strongly that the electrical structure of one is distorted by the presence
of the other. Knowing that both particles are in a sense composite (the
proton consisting of three quarks, the pion of two), we might expect that
to happen during a close approach. In fact, nothing much happens down
to a distance of 10 -15 m. At shorter distances, for strongly interacting
particles like the proton and the pion, nonelectrical forces dominate the
scene anyway.
That explains why we do not need to include “self-energy” terms
like the first two integrals in Eq. (1.56) in our energy accounts for a sys¬
tem of elementary charged particles. Indeed, we want to omit them. We
are doing just that, in effect, when we replace the actual distribution of
discrete elementary charges (the electrons on the rubber balloon) by a
perfectly continuous charge distribution.
1.16 Applications
Each chapter of this book concludes with a list of “everyday” applications
of the topics covered in the chapter. The discussions are brief. It would
take many pages to explain each item in detail; real-life physics tends to
involve countless variations, complications, and subtleties. The main pur¬
pose here is just to say a few words to convince you that the applications
are interesting and worthy of further study. You can carry onward with
some combination of books/internet/people/pondering. There is effec¬
tively an infinite amount of information out there, so you should take
advantage of it! Two books packed full of real-life applications are:
• The Flying Circus of Physics (Walker, 2007);
• How Things Work (Bloomfield, 2010).
And some very informative websites are:
• The Flying Circus of Physics website: www.flyingcircusofphysics.com;
• How Stuff Works: www.howstuffworks.com;
36
Electrostatics: charges and fields
• Explain That Stuff, www.explainthatstuff.com;
• and Wikipedia, of course: www.wikipedia.org.
These websites can point you to more technical sources if you want to
pursue things at a more advanced level.
With the exception of the gravitational force keeping us on the earth, and
ignoring magnets for the time being, essentially all “everyday” forces are
electrostatic in origin (with some quantum mechanics mixed in, to make
things stable; see Earnshaw’s theorem in Section 2.12). Friction, tension,
normal force, etc., all boil down to the electric forces between the elec¬
trons in the various atoms and molecules. You can open a door by push¬
ing on it because the forces between neighboring molecules in the door,
and also in your hand, are sufficiently strong. We can ignore the gravi¬
tational force between everyday-sized objects because the gravitational
force is so much weaker than the electric force (see Problem 1.1). Only
if one of the objects is the earth does the gravitational force matter. And
even in that case, it is quite remarkable that the electric forces between
the molecules in, say, a wooden board that you might be standing on
can completely balance the gravitational force on you due to the entire
earth. However, this wouldn’t be the case if you attempt to stand on a
lake (unless it’s frozen!).
If you want to give an object a net charge, a possible way is via the
triboelectric effect. If certain materials are rubbed against each other,
they can become charged. For example, rubbing wool and Teflon together
causes the wool to become positively charged and the Teflon negatively
charged. The mechanism is simple: the Teflon simply grabs electrons from
the wool. The determination of which material ends up with extra electrons
depends on the electronic structure of the molecules in the materials.
It turns out that actual rubbing isn’t necessary. Simply touching and
separating the materials can produce an imbalance of charge. Triboelectric
effects are mitigated by humid air, because the water molecules in the
air are inclined to give or receive electrons, depending on which of these
actions neutralizes the object. This is due to the fact that water molecules
are polar, that is, they are electrically lopsided. (Polar molecules will be
discussed in Chapter 10.)
The electrical breakdown of air occurs when the electric field
reaches a strength of about 3 • 10 6 V/m. In fields this strong, electrons
are ripped from molecules in the air. They are then accelerated by the
field and collide with other molecules, knocking electrons out of these
molecules, and so on, in a cascading process. The result is a spark,
because eventually the electrons will combine in a more friendly man¬
ner with molecules and drop down to a lower energy level, emitting the
light that you see. If you shuffle your feet on a carpet and then bring your
finger close to a grounded object, you will see a spark.
The electric field near the surface of the earth is about 100 V/m,
pointing downward. You can show that this implies a charge of —5 • 10 5 C
1.16 Applications
37
on the earth. The atmosphere contains roughly the opposite charge, so
that the earth-plus-atmosphere system is essentially neutral, as it must be.
(Why?) If there were no regenerative process, charge would leak between
the ground and the atmosphere, and they would neutralize each other in
about an hour. But there is a regenerative process: lightning. This is a
spectacular example of electrical breakdown. There are millions of light¬
ning strikes per day over the surface of the earth, the vast majority of
which transfer negative charge to the earth. A lightning strike is the result
of the strong electric field that is produced by the buildup (or rather, the
separation) of charge in a cloud. This separation arises from the charge
carried on moving raindrops, although the exact process is rather com¬
plicated (see the interesting discussion in Chapter 9 of Feynman el al.
(1977)). “Lightning” can also arise from the charge carried on dust parti¬
cles in coal mines, flour mills, grain storage facilities, etc. The result can
be a deadly explosion.
A more gentle form of electrical breakdown is corona discharge.
Near the tip of a charged pointy object, such as a needle, the field is large
but then falls off rapidly. (You can model the needle roughly as having
a tiny charged sphere on its end.) Electrons are ripped off the needle (or
off the air molecules) very close to the needle, but the field farther away
isn’t large enough to sustain the breakdown. So there is a slow leakage
instead of an abrupt spark. This leakage can sometimes be seen as a faint
glow. Examples are St. Elmo’s fire at the tips of ship masts, and a glow at
the tips of airplane wings.
Electrostatic paint sprayers can produce very even coats of paint. As
the paint leaves the sprayer, an electrode gives it a charge. This causes
the droplets in the paint mist to repel each other, helping to create a uni¬
form mist with no clumping. If the object being painted is grounded (or
given the opposite charge), the paint will be attracted to it, leading to less
wasted paint, less mess, and less inhalation of paint mist. When painting
a metal pipe, for example, the mist will wrap around and partially coat
the back side, instead of just sailing off into the air.
Photocopiers work by giving the toner powder a charge, and giving
certain locations on a drum or belt the opposite charge. These locations
on the drum can be made to correspond to the locations of ink on the
original paper. This is accomplished by coating the drum with a photo-
conductive material, that is, one that becomes conductive when exposed
to light. The entire surface of the drum is given an initial charge and
then exposed to light at locations corresponding to the white areas on the
original page (accomplished by reflecting light off the page). The charge
can be made to flow off these newly conductive locations on the drum,
leaving charge only at the locations corresponding to the ink. When the
oppositely charged toner is brought nearby, it is attracted to these loca¬
tions on the drum. The toner is then transferred to a piece of paper, pro¬
ducing the desired copy.
Electronic paper , used in many eBook readers, works by using
electric fields to rotate or translate small black and white objects.
38
Electrostatics: charges and fields
One technique uses tiny spheres (about 10 -4 m in diameter) that are
black on one side and white on the other, with the sides being oppo¬
sitely charged. Another technique uses similarly tiny spheres that are
filled with many even tinier charged white particles along with a dark
dye. In both cases, a narrow gap between sheets of electrodes (with one
sheet being the transparent sheet that you look through) is filled with
the spheres. By depositing a specific pattern of charge on the sheets, the
color of the objects facing your eye can be controlled. In the first system,
the black and white spheres rotate accordingly. In the second system, the
tiny white particles pile up on one side of the sphere. In contrast with a
standard LCD computer screen, electronic paper acts like normal paper,
in that it doesn’t produce its own light; an outside light source is needed
to view the page. An important advantage of electronic paper is that it
uses a very small amount of power. A battery is needed only when the
page is refreshed, whereas an LCD screen requires continual refreshing.
CHAPTER SUMMARY
• Electric charge, which can be positive or negative, is both conserved
and quantized. The force between two charges is given by Coulomb’s
law:
1 q\qiV2\
4:reo r \ x
(1.57)
Integrating this force, we find that the potential energy of a system of
charges (the work necessary to bring them in from infinity) equals
1 N
qm
2 J tr« 4 ’ ie » r *
(1.58)
• The electric field due to a charge distribution is (depending on whether
the distribution is continuous or discrete)
E = —— f
47T6 0 J
p(x',y',z!)rdx J dy'dz'
or
( 1 - 59 )
47reo “ rf
J= 1 J
The force on a test charge q due to the field is F = q E.
• The^?i« of an electric field through a surface S is
O = J E da.
(1.60)
Gauss’s law states that the flux of the electric field E through any
closed surface equals 1 /eo times the total charge enclosed by the
Problems
39
surface. That is (depending on whether the distribution is continuous
or discrete),
j E-da
pdv
(1.61)
Gauss’s law gives the fields for a sphere, line, and sheet of charge as
_ Q _ X _ cr
^sphere = , n, E\ i ne = , , Esheet = z ■ (1-62)
4jreor z 2eo
More generally, the discontinuity in the normal component of E across
a sheet is AEj_=<j/eo. Gauss’s law is always valid, although it is
useful for calculating the electric field only in cases where there is
sufficient symmetry.
• The force per unit area on a layer of charge equals the density times
the average of the fields on either side:
j= 1 -(E x +E 2 )o. (1.63)
• The energy density of an electric field is €qE 2 /2, so the total energy in
a system equals
U = -^ J E 2 dv. (1.64)
Problems
1.1 Gravity vs. electricity *
(a) In the domain of elementary particles, a natural unit of mass
is the mass of a nucleon, that is, a proton or a neutron, the
basic massive building blocks of ordinary matter. Given the
nucleon mass as 1.67 • 10 -27 kg and the gravitational constant
G as 6.67 • 10“ 11 m 3 /(kg s 2 ), compare the gravitational attrac¬
tion of two protons with their electrostatic repulsion. This
shows why we call gravitation a very weak force.
(b) The distance between the two protons in the helium nucleus
could be at one instant as much as 10“ 15 m. How large is the
force of electrical repulsion between two protons at that dis¬
tance? Express it in newtons, and in pounds. Even stronger is
the nuclear force that acts between any pair of hadrons (includ¬
ing neutrons and protons) when they are that close together.
1.2 Zero force from a triangle **
Two positive ions and one negative ion are fixed at the vertices of
an equilateral triangle. Where can a fourth ion be placed, along the
symmetry axis of the setup, so that the force on it will be zero? Is
there more than one such place? You will need to solve something
numerically.
Electrostatics: charges and fields
Q
<?
d
1.3 Force from a cone **
(a) A charge q is located at the tip of a hollow cone (such as an ice
cream cone without the ice cream) with surface charge density
a. The slant height of the cone is L, and the half-angle at the
vertex is 0. What can you say about the force on the charge q
due to the cone?
(b) If the top half of the cone is removed and thrown away (see
Fig. 1.31), what is the force on the charge q due to the remain¬
ing part of the cone? For what angle 6 is this force maximum?
1.4 Work for a rectangle **
Two protons and two electrons are located at the corners of a rect¬
angle with side lengths a and b. There are two essentially different
arrangements. Consider the work required to assemble the system,
starting with the particles very far apart. Is it possible for the work
to be positive for either of the arrangements? If so, how must a and
b be related? You will need to solve something numerically.
1.5 Stable or unstable? **
In the setup in Exercise 1.37, is the charge — Q at the center of the
square in stable or unstable equilibrium? You can answer this by
working with either forces or energies. The latter has the advan¬
tage of not involving components, although things can still get
quite messy. However, the math is simple if you use a computer.
Imagine moving the —Q charge infinitesimally to the point (x,y),
and use, for example, the Series operation in Matliematica to cal¬
culate the new energy of the charge, to lowest nontrivial order in
x and y. If the energy decreases for at least one direction of dis¬
placement, then the equilibrium is unstable. (The equilibrium is
certainly stable with respect to displacements perpendicular to the
plane of the square, because the attractive force from the other
charges is directed back toward the plane. The question is, what
happens in the plane of the square?)
1.6 Zero potential energy for equilibrium **
(a) Two charges q are each located a distance d from a charge Q,
as shown in Fig. 1.32(a). What should the charge Q be so that
the system is in equilibrium; that is, so that the force on each
charge is zero? (The equilibrium is an unstable one, which can
be seen by looking at longitudinal displacements of the (nega¬
tive) charge Q. This is consistent with a general result that we
will derive Section 2.12.)
(b) Same question, but now with the setup in Fig. 1.32(b). The
three charges q are located at the vertices of an equilateral
triangle.
(c) Show that the total potential energy in each of the above
systems is zero.
Problems
41
(d) In view of the previous result, we might make the follow¬
ing conjecture: “The total potential energy of any system of
charges in equilibrium is zero.” Prove that this conjecture is
indeed true. Hint: The goal is to show that zero work is required
to move the charges out to infinity. Since the electrostatic force
is conservative, you need only show that the work is zero for
one particular set of paths of the charges. And there is indeed
a particular set of paths that makes the result clear.
1.7 Potential energy in a two-dimensional crystal **
Use a computer to calculate numerically the potential energy, per
ion, for an infinite two-dimensional square ionic crystal with sepa¬
ration a; that is, a plane of equally spaced charges of magnitude e
and alternating sign (as with a checkerboard).
1.8 Oscillating in a ring ***
A ring with radius R has uniform positive charge density /.. A par¬
ticle with positive charge q and mass m is initially located at the
center of the ring and is then given a tiny kick. If it is constrained
to move in the plane of the ring, show that it undergoes simple
harmonic motion (for small oscillations), and find the frequency.
Hint: Find the potential energy of the particle when it is at a (small)
radius, r, by integrating over the ring, and then take the negative
derivative to find the force. You will need to use the law of cosines
and also the Taylor series l/y/\ + e ^ 1 — e/2 + 3e 2 /8.
1.9 Field from two charges **
A charge 2 q is at the origin, and a charge —q is at x = a on the x
axis.
(a) Find the point on the x axis where the electric field is zero.
(b) Consider the vertical line passing through the charge —q, that
is, the line given by x = a. Locate, at least approximately, a
point on this line where the electric field is parallel to the x «
axis. |
I
I
1.10 45-degree field line ** \l
A half-infinite line has linear charge density X. Find the electric |
field at a point that is “even” with the end, a distance l from it, as A !
shown in Fig. 1.33. You should find that the field always points up
at a 45° angle, independent of i. Figure 1.33.
1.11 Field at the end of a cylinder **
(a) Consider a half-infinite hollow cylindrical shell (that is, one
that extends to infinity in one direction) with radius R and uni¬
form surface charge density a. What is the electric field at the
midpoint of the end face?
(b) Use your result to determine the field at the midpoint of a
half-infinite solid cylinder with radius R and uniform volume
42
Electrostatics: charges and fields
z
-Q
h
Q
Figure 1.35.
charge density p, which can be considered to be built up from
many cylindrical shells.
1.12 Field from a hemispherical shell ***
A hemispherical shell has radius R and uniform surface charge
density a (see Fig. 1.34). Find the electric field at a point on the
symmetry axis, at position z relative to the center, for any z value
from — oo to oo.
1.13 A very uniform field ***
(a) Two rings with radius r have charge Q and —Q uniformly
distributed around them. The rings are parallel and located a
distance li apart, as shown in Fig. 1.35. Let z be the vertical
coordinate, with z = 0 taken to be at the center of the lower
ring. As a function of z, what is the electric field at points on
the axis of the rings?
(b) You should find that the electric field is an even function with
respect to the z = h/2 point midway between the rings. This
implies that, at this point, the held has a local extremum as a
function of z. The held is therefore fairly uniform there; there
are no variations to first order in the distance along the axis
from the midpoint. What should r be in terms of li so that the
held is very uniform?
By “very” uniform we mean that additionally there aren’t
any variations to second order in z. That is, the second deriva¬
tive vanishes. This then implies that the leading-order change
is fourth order in z (because there are no variations at any odd
order, since the held is an even function around the midpoint).
Feel free to calculate the derivatives with a computer.
1.14 Hole in a plane **
(a) A hole of radius R is cut out from a very large Hat sheet with
uniform charge density a. Let L be the line perpendicular to
the sheet, passing through the center of the hole. What is the
electric held at a point on L, a distance z from the center of the
hole? Hint: Consider the plane to consist of many concentric
rings.
(b) If a charge —q with mass m is released from rest on L, very
close to the center of the hole, show that it undergoes oscil¬
latory motion, and hnd the frequency co of these oscillations.
What is co if m = 1 g, —q = —10 -8 C, a = 10 -6 C/m 2 , and
R = 0.1 m?
(c) If a charge — q with mass m is released from rest on L, a dis¬
tance z from the sheet, what is its speed when it passes through
the center of the hole? What does your answer reduce to for
large z (or, equivalently, small R)1
Problems
43
1.15 Flux through a circle **
A point charge q is located at the origin. Consider the electric field
flux through a circle a distance l from q, subtending an angle 20, as
shown in Fig. 1.36. Since there are no charges except at the origin,
any surface that is bounded by the circle and that stays to the right
of the origin must contain the same flux. (Why?) Calculate this
flux by taking the surface to be:
(a) the flat disk bounded by the circle;
(b) the spherical cap (with the sphere centered at the origin)
bounded by the circle.
"-V
Figure 1.36.
1.16 Gauss’s law and two point charges **
(a) Two point charges q are located at positions x = ±£. At points
close to the origin on the x axis, find E x . At points close to the
origin on the y axis, find E y . Make suitable approximations
with x <£l and y l.
(b) Consider a small cylinder centered at the origin, with its axis
along the jr axis. The radius is ro and the length is 2 xq. Using
your results from part (a), verify that there is zero flux through
the cylinder, as required by Gauss’s law.
1.17 Zero field inside a spherical shell **
Consider a hollow spherical shell with uniform surface charge den¬
sity. By considering the two small patches at the ends of the thin
cones in Fig. 1.37, show that the electric field at any point P in
the interior of the shell is zero. This then implies that the electric
potential (defined in Chapter 2) is constant throughout the interior.
1.18 Fields at the surfaces **
Consider the electric field at a point on the surface of (a) a sphere
with radius R, (b) a cylinder with radius R whose length is infinite,
and (c) a slab with thickness 2 R whose other two dimensions are
infinite. All of the objects have the same volume charge density p.
Compare the fields in the three cases, and explain physically why
the sizes take the order they do.
1.19 Sheet on a sphere **
Consider a large flat horizontal sheet with thickness x and volume
charge density p. This sheet is tangent to a sphere with radius R
and volume charge density po, as shown in Fig. 1.38. Let A be the
point of tangency, and let B be the point opposite to A on the top
side of the sheet. Show that the net upward electric field (from
the sphere plus the sheet) at B is larger than at A if p > (2/3)p 0 .
(Assume x <SC R.)
B P
44
Electrostatics: charges and fields
1.20 Thundercloud **
You observe that the passage of a particular thundercloud over¬
head causes the vertical electric field strength in the atmosphere,
measured at the ground, to rise to 3000 N/C (or V/m).
(a) How much charge does the thundercloud contain, in coulombs
per square meter of horizontal area? Assume that the width of
the cloud is large compared with the height above the ground.
(b) Suppose there is enough water in the thundercloud in the form
of 1 mm diameter drops to make 0.25 cm of rainfall, and that
it is those drops that carry the charge. How large is the electric
field strength at the surface of one of the drops?
1.21 Field in the end face *
Consider a half-infinite hollow cylindrical shell (that is, one that
extends to infinity in one direction) with uniform surface charge
density. Show that at all points in the circular end face, the elec¬
tric field is parallel to the cylinder’s axis. Hint: Use superposition,
along with what you know about the field from an infinite (in both
directions) hollow cylinder.
1.22 Field from a spherical shell, right and wrong **
The electric field outside and an infinitesimal distance away from a
uniformly charged spherical shell, with radius R and surface charge
density cr, is given by Eq. (1.42) as a/e o. Derive this in the follow¬
ing way.
(a) Slice the shell into rings (symmetrically located with respect to
the point in question), and then integrate the field contributions
from all the rings. You should obtain the incorrect result of
a/2e 0 .
(b) Why isn’t the result correct? Explain how to modify it to obtain
the correct result of o/e o- Hint: You could very well have per¬
formed the above integral in an effort to obtain the electric
field an infinitesimal distance inside the shell, where we know
the field is zero. Does the above integration provide a good
description of what’s going on for points on the shell that are
very close to the point in question?
1.23 Field near a stick **
A stick with length 21 has uniform linear charge density X. Con¬
sider a point P , a distance r)t from the center (where 0 < p < 1),
and an infinitesimal distance away from the stick. Up close, the
stick looks infinitely long, as far as the E component perpendicu¬
lar to the stick is concerned. So we have E± = X/2jTCQr. Find the
E component parallel to the stick, E\\. Does it approach infinity, or
does it remain finite at the end of the stick?
Problems
45
1.24 Potential energy of a cylinder ***
A cylindrical volume of radius a is filled with charge of uniform
density p. We want to know the potential energy per unit length
of this cylinder of charge, that is, the work done per unit length
in assembling it. Calculate this by building up the cylinder layer
by layer, making use of the fact that the field outside a cylindrical
distribution of charge is the same as if all the charge were located
on the axis. You will find that the energy per unit length is infinite
if the charges are brought in from infinity, so instead assume that
they are initially distributed uniformly over a hollow cylinder with
large radius R. Write your answer in terms of the charge per unit
length of the cylinder, which is A. = pica 2 . (See Exercise 1.83 for
a different method of solving this problem.)
1.25 Two equal fields **
The result of Exercise 1.78 is that the electric field at the center
of a small hole in a spherical shell equals a/2 cq. This happens
to be the same as the field due to an infinite flat sheet with the
same density cr. That is, at the center of the hole at the top of the
spherical shell in Fig. 1.39, the field from the shell equals the field
from the infinite horizontal sheet shown. (This sheet could actually
be located at any height.) Demonstrate this equality by explaining
why the rings on the shell and sheet that are associated with the
angle 0 and angular width dO yield the same field at the top of the
shell.
Figure 1.39.
1.26 Stable equilibrium in electron jelly **
The task of Exercise 1.77 is to find the equilibrium positions of
two protons located inside a sphere of electron jelly with total
charge — 2e. Show that the equilibria are stable. That is, show that
a displacement in any direction will result in a force directed back
toward the equilibrium position. (There is no need to know the
exact locations of the equilibria, so you can solve this problem
without solving Exercise 1.77 first.)
1.27 Uniform field in a cavity **
A sphere has radius R\ and uniform volume charge density p. A
spherical cavity with radius Ri is carved out at an arbitrary loca¬
tion inside the larger sphere. Show that the electric field inside the
cavity is uniform (in both magnitude and direction). Hint: Find a
vector expression for the field in the interior of a charged sphere,
and then use superposition.
What are the analogous statements for the lower-dimensional
analogs with cylinders and slabs? Are the statements still true?
1.28 Average field on/in a sphere **
(a) A point charge q is located at an arbitrary position inside a
sphere (just an imaginary sphere in space) with radius R. Show
46
Electrostatics: charges and fields
that the average electric field over the surface of the sphere
is zero. Hint: Use an argument involving Newton’s third law,
along with what you know about spherical shells.
(b) If the point charge q is instead located outside the sphere, a
distance r from the center, show that the average electric field
over the surface of the sphere has magnitude q/4jteor.
(c) Return to the case where the point charge q is located inside
the sphere of radius R. Let the distance from the center be r.
Use the above results to show that the average electric field
over the entire volume of the sphere of radius R has magnitude
qr/4-neoR 3 and points toward the center (if q is positive).
1.29 Pulling two sheets apart **
Two parallel sheets each have large area A and are separated by
a small distance l. The surface charge densities are a and —a.
You wish to pull one of the sheets away from the other, by a small
distance x. How much work does this require? Calculate this by:
(a) using the relation W = (force) x (distance);
(b) calculating the increase in energy stored in the electric field.
Show that these two methods give the same result.
1.30 Force on a patch **
Consider a small patch of charge that is part of a larger surface.
The surface charge density is a. If Ei and E 2 are the electric fields
on either side of the patch, show that the force per unit area on
the patch equals cr(Ei + E2)/2. This is the result we derived in
Section 1.14, for the case where the field is perpendicular to the
surface. Derive it here by using the fact that the force on the patch
is due to the field E other from all the other charges in the system
(excluding the patch), and then finding an expression for E other in
terms of Ei and Et.
1.31 Decreasing energy? *
A hollow spherical shell with radius R has charge Q uniformly dis¬
tributed over it. The task of Problem 1.32 is to show that the energy
stored in this system is Q 2 /8jteoR. (You can derive this here if you
want, or you can just accept it for the purposes of this problem.)
Now imagine taking all of the charge and concentrating it in two
point charges Q/ 2 located at diametrically opposite positions on
the shell. The energy of this new system is (Q/2) 2 /4tc eo(2R) =
Q 2 /32 ttcoR, which is less than the energy of the uniform spheri¬
cal shell. Does this make sense? If not, where is the error in this
reasoning?
Exercises
47
1.32 Energy of a shell **
A hollow spherical shell with radius R has charge Q uniformly
distributed over it. Show that the energy stored in this system is
Q 2 /SneoR. Do this in two ways as follows.
(a) Use Eq. (1.53) to find the energy stored in the electric field.
(b) Imagine building up the shell by successively adding on
infinitesimally thin shells with charge dq. Find the energy
needed to add on a shell when the charge already there is q,
and then integrate over q.
1.33 Deriving the energy density ***
Consider the electric field of two protons a distance b apart. Accord¬
ing to Eq. (1.53) (which we stated but did not prove), the
potential energy of the system ought to be given by
where Ei is the field of one particle alone and Et that of the other.
The first of the three integrals on the right might be called the
“electrical self-energy” of one proton; an intrinsic property of the
particle, it depends on the proton’s size and structure. We have
always disregarded it in reckoning the potential energy of a sys¬
tem of charges, on the assumption that it remains constant; the
same goes for the second integral. The third integral involves the
distance between the charges. Evaluate this integral. This is most
easily done if you set it up in spherical polar coordinates with one
of the protons at the origin and the other on the polar axis, and
perform the integration over r before the integration over 0. Thus,
by direct calculation, you can show that the third integral has the
value e 2 /47T6o b, which we already know to be the work required to
bring the two protons in from an infinite distance to positions a dis¬
tance b apart. So you will have proved the correctness of Eq. (1.53)
for this case, and by invoking superposition you can argue that
Eq. (1.53) must then give the energy required to assemble any sys¬
tem of charges.
Exercises
1.34 Aircraft carriers and specks of gold *
Imagine (quite unrealistically) removing one electron from
every atom in a tiny cube of gold 1 mm on a side. (Never mind how
you would hold the resulting positively charged cube together.) Do
the same thing with another such cube a meter away. What is the
repulsive force between the two cubes? How many aircraft carriers
2.5 m
48
Electrostatics: charges and fields
Figure 1.40.
would you need in order to have their total weight equal this force?
Some data: The density of gold is 19.3 g/cm 3 , and its molecular
weight is 197; that is, 1 mole (6.02 • 10 23 ) of gold atoms has a mass
of 197 grams. The mass of an aircraft carrier is around 100 million
kilograms.
1.35 Balancing the weight *
On the utterly unrealistic assumption that there are no other charged
particles in the vicinity, at what distance below a proton would the
upward force on an electron equal the electron’s weight? The mass
of an electron is about 9 • 10 -31 kg.
1.36 Repelling volley balls *
Two volley balls, mass 0.3 kg each, tethered by nylon strings and
charged with an electrostatic generator, hang as shown in
Fig. 1.40. What is the charge on each, assuming the charges are
equal?
1.37 Zero force at the corners **
(a) At each corner of a square is a particle with charge q. Fixed at
the center of the square is a point charge of opposite sign, of
magnitude Q. What value must Q have to make the total force
on each of the four particles zero?
(b) With Q taking on the value you just found, show that the poten¬
tial energy of the system is zero, consistent with the result from
Problem 1.6.
1.38 Oscillating on a line **
Two positive point charges Q are located at points (±£, 0). A par¬
ticle with positive charge q and mass m is initially located midway
between them and is then given a tiny kick. If it is constrained to
move along the line joining the two charges Q, show that it under¬
goes simple harmonic motion (for small oscillations), and find the
frequency.
1.39 Rhombus of charges **
Four positively charged bodies, two with charge Q and two with
charge q, are connected by four unstretchable strings of equal length.
In the absence of external forces they assume the equilibrium
configuration shown in Fig. 1.41. Show that tan 3 0 = q 1 /Q 1 . This
can be done in two ways. You could show that this relation must hold
if the total force on each body, the vector sum of string tension and
electrical repulsion, is zero. Or you could write out the expression
for the energy U of the assembly (like Eq. (1.13) but for four charges
instead of three) and minimize it.
1.40 Zero potential energy **
Find a geometrical arrangement of one proton and two electrons
such that the potential energy of the system is exactly zero. How
Exercises
49
many such arrangements are there with the three particles on the
same straight line? You should find that the ratio of two of the
distances involved is the golden ratio.
1.41 Work for an octahedron **
Three protons and three electrons are to be placed at the vertices of
a regular octahedron of edge length a. We want to find the energy
of the system, that is, the work required to assemble it starting
with the particles very far apart. There are two essentially different
arrangements. What is the energy of each?
1.42 Potential energy in a one-dimensional crystal **
Calculate the potential energy, per ion, for an infinite ID ionic
crystal with separation a; that is, a row of equally spaced charges
of magnitude e and alternating sign. Hint: The power-series expan¬
sion of ln(l + x ) may be of use.
1.43 Potential energy in a three-dimensional crystal **
In the spirit of Problem 1.7, use a computer to calculate numeric¬
ally the potential energy, per ion, for an infinite 3D cubic ionic
crystal with separation a. In other words, derive Eq. (1.18).
1.44 Chessboard **
An infinite chessboard with squares of side s has a charge e at
the center of every white square and a charge — e at the center of
every black square. We are interested in the work W required to
transport one charge from its position on the board to an infinite
distance from the board. Given that W is finite (which is plausible
but not so easy to prove), do you think it is positive or negative?
Calculate an approximate value for W by removing the charge from
the central square of a 7 x 7 board. (Only nine different terms are
involved in that sum.) For larger arrays you can write a program
to compute the work numerically. This will give you some idea of
the rate of convergence toward the value for the infinite array; see
Problem 1.7.
1.45 Zerofield? **
Four charges, q, —q, q, and —q, are located at equally spaced inter¬
vals on the x axis. Their x values are —3a, —a, a, and 3a, respec¬
tively. Does there exist a point on the y axis for which the electric
field is zero? If so, find the y value.
1.46 Charges on a circular track **
Suppose three positively charged particles are constrained to move
on a fixed circular track. If the charges were all equal, an equi¬
librium arrangement would obviously be a symmetrical one with
the particles spaced 120° apart around the circle. Suppose that two
50
Electrostatics: charges and fields
Figure 1.42.
of the charges are equal and the equilibrium arrangement is such
that these two charges are 90° apart rather than 120°. What is the
relative magnitude of the third charge?
1.47 Field from a semicircle *
A thin plastic rod bent into a semicircle of radius R has a charge Q
distributed uniformly over its length. Find the electric field at the
center of the semicircle.
1.48 Maximum field from a ring **
A charge Q is distributed uniformly around a thin ring of radius
b that lies in the xy plane with its center at the origin. Locate the
point on the positive z axis where the electric field is
strongest.
1.49 Maximum field from a blob **
(a) A point charge is placed somewhere on the curve shown in
Fig. 1.42. This point charge creates an electric field at the ori¬
gin. Let E y be the vertical component of this field. What shape
(up to a scaling factor) should the curve take so that E y is inde¬
pendent of the position of the point charge on the curve?
(b) You have a moldable material with uniform volume charge
density. What shape should the material take if you want to
create the largest possible electric field at a given point in
space? Be sure to explain your reasoning clearly.
1.50 Field from a hemisphere **
(a) What is the electric field at the center of a hollow hemispheri¬
cal shell with radius R and uniform surface charge density er?
(This is a special case of Problem 1.12, but you can solve the
present exercise much more easily from scratch, without going
through all the messy integrals of Problem 1.12.)
(b) Use your result to show that the electric field at the center of
a solid hemisphere with radius R and uniform volume charge
density p equals pR/4eo-
1.51 N charges on a circle ***
N point charges, each with charge Q/N, are evenly distributed
around a circle of radius R. What is the electric field at the loca¬
tion of one of the charges, due to all the others? (You can leave
your answer in the form of a sum.) In the N oo limit, is the
field infinite or finite? In the N —»■ oo limit, is the force on one of
the charges infinite or finite?
1.52 An equilateral triangle *
Three positive charges, A , B, and C, of 3 • 10 -6 , 2 • 10 -6 , and
2 ■ 10 -6 coulombs, respectively, are located at the corners of an
equilateral triangle of side 0.2 m.
Exercises
51
(a) Find the magnitude in newtons of the force on each charge.
(b) Find the magnitude in newtons/coulomb of the electric field at
the center of the triangle.
1.53 Concurrent field lines **
A semicircular wire with radius R has uniform charge density —X.
Show that at all points along the “axis” of the semicircle (the line
through the center, perpendicular to the plane of the semicircle, as
shown in Fig. 1.43), the vectors of the electric field all point toward
a common point in the plane of the semicircle. Where is this point?
1.54 Semicircle and wires **
(a) Two long, thin parallel rods, a distance 2b apart, are joined by a
semicircular piece of radius b, as shown in Fig. 1.44. Charge of
uniform linear density X is deposited along the whole filament.
Show that the field E of this charge distribution vanishes at the
point C. Do this by comparing the contribution of the element
at A to that of the element at B which is defined by the same
values of 9 and dd.
(b) Consider the analogous two-dimensional setup involving a
cylinder and a hemispherical end cap, with uniform surface
charge density o. Using the result from part (a), do you think
that the field at the analogous point C is directed upward,
downward, or is zero? (No calculations needed!)
1.55 Field from a finite rod **
A thin rod 10 cm long carries a total charge of 24 esu = 8 • 10~ 9 C
uniformly distributed along its length. Find the strength of the elec¬
tric field at each of the two points A and B located as shown in
Fig. 1.45.
1.56 Flux through a cube *
(a) A point charge q is located at the center of a cube of edge d.
What is the value of f E • da over one face of the cube?
(b) The charge q is moved to one corner of the cube. Now what
is the value of the flux of E through each of the faces of the
cube? (To make things well defined, treat the charge like a tiny
sphere.)
1.57 Escaping field lines **
Charges 2 q and —q are located on the x axis at x — 0 and x = a,
respectively.
(a) Find the point on the x axis where the electric field is zero, and
make a rough sketch of some field lines.
(b) You should find that some of the field lines that start on the
2 q charge end up on the —q charge, while others head off to
infinity. Consider the field lines that form the cutoff between
these two cases. At what angle (with respect to the x axis) do
axis
*
52
Electrostatics: charges and fields
h:
3 cm
10 cm
5 cm
B •
Figure 1.45.
these lines leave the 2 q charge? Hint: Draw a wisely chosen
Gaussian surface that mainly follows these lines.
1.58 Gauss’s law at the center of a ring **
(a) A ring with radius R has total charge Q uniformly distributed
around it. To leading order, find the electric held at a point
along the axis of the ring, a very small distance z from the
center.
(b) Consider a small cylinder centered at the center of the ring,
with small radius ro and small height 2zo, with zo lying on
either side of the plane of the ring. There is no charge in this
cylinder, so the net flux through it must be zero. Using a result
given in the solution to Problem 1.8, verify that this is indeed
the case (to leading order in the small distances involved).
1.59 Zero field inside a cylindrical shell *
Consider a distribution of charge in the form of a hollow circular
cylinder, like a long charged pipe. In the spirit of Problem 1.17,
show that the electric field inside the pipe is zero.
1.60 Field from a hollow cylinder *
Consider the hollow cylinder from Exercise 1.59. Use Gauss’s law
to show that the field inside the pipe is zero. Also show that the
field outside is the same as if the charge were all on the axis. Is
either statement true for a pipe of square cross section on which
the charge is distributed with uniform surface density?
1.61 Potential energy of a sphere **
A spherical volume of radius R is filled with charge of uniform
density p. We want to know the potential energy U of this sphere
of charge, that is, the work done in assembling it. In the example
in Section 1.15, we calculated U by integrating the energy density
of the electric field; the result was U = {3 / 5)Q 2 / An cqR. Derive
U here by building up the sphere layer by layer, making use of the
fact that the field outside a spherical distribution of charge is the
same as if all the charge were at the center.
1.62 Electron self-energy *
At the beginning of the twentieth century the idea that the rest
mass of the electron might have a purely electrical origin was very
attractive, especially when the equivalence of energy and mass
was revealed by special relativity. Imagine the electron as a ball of
charge, of constant volume density out to some maximum radius
ro- Using the result of Exercise 1.61, set the potential energy of
this system equal to me 2 and see what you get for ro- One defect of
the model is rather obvious: nothing is provided to hold the charge
together!
Exercises
53
1.63 Sphere and cones **
(a) Consider a fixed hollow spherical shell with radius R and sur¬
face charge density a. A particle with mass m and charge —q
that is initially at rest falls in from infinity. What is its speed
when it reaches the center of the shell? (Assume that a tiny
hole has been cut in the shell, to let the charge through.)
(b) Consider two fixed hollow conical shells (that is, ice cream
cones without the ice cream) with base radius R, slant height L,
and surface charge density er, arranged as shown in Fig. 1.46.
A particle with mass m and charge —q that is initially at rest
falls in from infinity, along the perpendicular bisector line, as
shown. What is its speed when it reaches the tip of the cones?
You should find that your answer relates very nicely to your
answer for part (a).
1.64 Field between two wires *
Consider a high-voltage direct current power line that consists of
two parallel conductors suspended 3 meters apart. The lines are
oppositely charged. If the electric field strength halfway between
them is 15,000 N/C, how much excess positive charge resides on a
1 km length of the positive conductor?
1.65 Building a sheet from rods **
An infinite uniform sheet of charge can be thought of as consisting
of an infinite number of adjacent uniformly charged rods. Using the
fact that the electric field from an infinite rod is X/2neor, integrate
over these rods to show that the field from an infinite sheet with
charge density o is <r/2eo-
1.66 Force between two strips **
(a) The two strips of charge shown in Fig. 1.47 have width b,
infinite height, and negligible thickness (in the direction per¬
pendicular to the page). Their charge densities per unit area are
±er. Find the magnitude of the electric field due to one of the
strips, a distance x away from it (in the plane of the page).
(b) Show that the force (per unit height) between the two strips
equals <r 2 b( In 2) /ueo. Note that this result is finite, even though
you will find that the field due to a strip diverges as you get
close to it.
1.67 Field from a cylindrical shell, right and wrong **
Find the electric field outside a uniformly charged hollow cylin¬
drical shell with radius R and charge density a , an infinitesimal
distance away from it. Do this in the following way.
(a) Slice the shell into parallel infinite rods, and integrate the field
contributions from all the rods. You should obtain the incor¬
rect result of ct/2€q.
Figure 1.46.
b b
a -a
Figure 1.47.
54
Electrostatics: charges and fields
Figure 1.48.
a
Figure 1.49.
(b) Why isn’t the result correct? Explain how to modify it to obtain
the correct result of cr/eo- Hint: You could very well have per¬
formed the above integral in an effort to obtain the electric
field an infinitesimal distance inside the cylinder, where we
know the field is zero. Does the above integration provide a
good description of what’s going on for points on the shell
that are very close to the point in question?
1.68 Uniform field strength *
We know from the example in Section 1.11 that the electric field
inside a solid sphere with uniform charge density is proportional
to r. Assume instead that the charge density is not uniform, but
depends only on r. What should this dependence be so that the
magnitude of the field at points inside the sphere is independent
of r (except right at the center, where it isn’t well defined)? What
should the dependence be in the analogous case where we have a
cylinder instead of a sphere?
1.69 Carved-out sphere **
A sphere of radius a is filled with positive charge with uniform
density p. Then a smaller sphere of radius a/ 2 is carved out, as
shown in Fig. 1.48, and left empty. What are the direction and mag¬
nitude of the electric field at A? At B1
1.70 Field from two sheets *
Two infinite plane sheets of surface charge, with densities 3cro and
—2ero, are located a distance i apart, parallel to one another. Dis¬
cuss the electric field of this system. Now suppose the two planes,
instead of being parallel, intersect at right angles. Show what the
field is like in each of the four regions into which space is thereby
divided.
1.71 Intersecting sheets **
(a) Figure 1.49 shows the cross section of three infinite sheets
intersecting at equal angles. The sheets all have surface charge
density a. By adding up the fields from the sheets, find the
electric field at all points in space.
(b) Find the field instead by using Gauss’s law. You should explain
clearly why Gauss’s law is in fact useful in this setup.
(c) What is the field in the analogous setup where there are N
sheets instead of three? What is your answer in the N -* oo
limit? This limit is related to the cylinder in Exercise 1.68.
1.72 A plane and a slab **
An infinite plane has uniform surface charge density a. Adjacent to
it is an infinite parallel layer of charge of thickness d and uniform
volume charge density p, as shown in Fig. 1.50. All charges are
fixed. Find E everywhere.
Figure 1.50.
Exercises
55
1.73 Sphere in a cylinder **
An infinite cylinder with uniform volume charge density p has its
axis lying along the z axis. A sphere is carved out of the cylin¬
der and then filled up with a material with uniform density —p/2.
Assume that the center of the sphere is located on the x axis at posi¬
tion x — a. Show that inside the sphere the component of the field
in the xy plane is uniform, and find its value. Hint: The technique
used in Problem 1.27 will be helpful.
1.74 Zero field in a sphere **
In Fig. 1.51 a sphere with radius R is centered at the origin, an
infinite cylinder with radius R has its axis along the z axis, and an
infinite slab with thickness 2 R lies between the planes z= — R and
z = R. The uniform volume densities of these objects are p \, pi,
and P3, respectively. The objects are superposed on top of each
other; the densities add where the objects overlap. How should
the three densities be related so that the electric field is zero
everywhere throughout the volume of the sphere? Hint: Find a
vector expression for the field inside each object, and then use
superposition.
1.75 Ball in a sphere **
We know that if a point charge q is located at radius a in the interior
of a sphere with radius R and uniform volume charge density p,
then the force on the point charge is effectively due only to the
charge that is located inside radius a.
P2
Cylinder
f R
_
\spf
R \
Pi
ere /
P 3
Slab
£. j;
Figure 1.51.
(a) Consider instead a uniform ball of charge located entirely inside
a larger sphere of radius R. Let the ball’s radius be h. and let
its center be located at radius a in the larger sphere. Its volume
charge density is such that its total charge is q. Assume that
the ball is superposed on top of the sphere, so that all of the
sphere’s charge is still present. Can the force on the ball be
obtained by treating it like a point charge and considering only
the charge in the larger sphere that is inside radius al
(b) Would the force change if we instead remove the charge in the
larger sphere where the ball is? So now we are looking at the
force on the ball due to the sphere with a cavity carved out,
which is a more realistic scenario.
1.76 Hydrogen atom **
The neutral hydrogen atom in its normal state behaves, in some
respects, like an electric charge distribution that consists of a point
charge of magnitude e surrounded by a distribution of negative
charge whose density is given by p(r) = —Ce~ 2r ^ a °. Here «o is
the Bohr radius, 0.53 ■ 10“ 10 m, and C is a constant with the value
required to make the total amount of negative charge exactly e.
56
Electrostatics: charges and fields
What is the net electric charge inside a sphere of radius ao ? What
is the electric field strength at this distance from the nucleus?
1.77 Electron jelly **
Imagine a sphere of radius a filled with negative charge of uniform
density, the total charge being equivalent to that of two electrons.
Imbed in this jelly of negative charge two protons, and assume that,
in spite of their presence, the negative charge distribution remains
uniform. Where must the protons be located so that the force on
each of them is zero? (This is a surprisingly realistic caricature of
a hydrogen molecule; the magic that keeps the electron cloud in
the molecule from collapsing around the protons is explained by
quantum mechanics!)
1.78 Hole in a shell **
Figure 1.52 shows a spherical shell of charge, of radius a and sur¬
face density er, from which a small circular piece of radius h <<C a
has been removed. What is the direction and magnitude of the field
at the midpoint of the aperture? There are two ways to get the
answer. You can integrate over the remaining charge distribution
to sum the contributions of all elements to the field at the point
in question. Or, remembering the superposition principle, you can
think about the effect of replacing the piece removed, which itself
is practically a little disk. Note the connection of this result with
our discussion of the force on a surface charge - perhaps that is a
third way in which you might arrive at the answer.
1.79 Forces on three sheets **
Consider three charged sheets. A, B, and C. The sheets are parallel
with A above B above C. On each sheet there is surface charge of
uniform density: -4 - 10“ 5 C/m 2 on A, 7- 10“ 5 C/m 2 on B, and
—3 ■ 10“ 5 C/m 2 on C. (The density given includes charge on both
sides of the sheet.) What is the magnitude of the electrical force
per unit area on each sheet? Check to see that the total force per
unit area on the three sheets is zero.
1.80 Force in a soap bubble **
Like the charged rubber balloon described at the end of Section 1.14,
a charged soap bubble experiences an outward electrical force on
every bit of its surface. Given the total charge Q on a bubble of
radius R, what is the magnitude of the resultant force tending to
pull any hemispherical half of the bubble away from the other half?
(Should this force divided by 2nR exceed the surface tension of
the soap film, interesting behavior might be expected!)
1.81 Energy around a sphere *
A sphere of radius R has a charge Q distributed uniformly over
its surface. How large a sphere contains 90 percent of the energy
stored in the electrostatic field of this charge distribution?
Exercises
57
1.82 Energy of concentric shells *
(a) Concentric spherical shells of radius a and b, with a < b,
carry charge Q and —Q, respectively, each charge uniformly
distributed. Find the energy stored in the electric field of this
system.
(b) Calculate the stored energy in a second way: start with two
neutral shells, and then gradually transfer positive charge from
the outer shell to the inner shell in a spherically symmetric
manner. At an intermediate stage when there is charge q on
the inner shell, find the work required to transfer an additional
charge dq. And then integrate over q.
1.83 Potential energy of a cylinder **
Problem 1.24 gives one way of calculating the energy per unit
length stored in a solid cylinder with radius a and uniform volume
charge density p. Calculate the energy here by using Eq. (1.53)
to find the total energy per unit length stored in the electric field.
Don’t forget to include the field inside the cylinder.
You will find that the energy is infinite, so instead calculate the
energy relative to the configuration where all the charge is initially
distributed uniformly over a hollow cylinder with large radius R.
(The field outside radius R is the same in both configurations, so
it can be ignored when calculating the relative energy.) In terms of
the total charge X per unit length in the final cylinder, show that the
energy per unit length can be written as fk 2 /Aneo) (1 /4+ln (R/af).
The electric
potential
Overview The first half of this chapter deals mainly with the
potential associated with an electric field. The second half covers
a number of mathematical topics that will be critical in our treat¬
ment of electromagnetism. The potential difference between two
points is defined to be the negative line integral of the electric field.
Equivalently, the electric field equals the negative gradient of the
potential. Just as the electric field is the force per unit charge, the
potential is the potential energy per unit charge. We give a num¬
ber of examples involving the calculation of the potential due to
a given charge distribution. One important example is the dipole,
which consists of two equal and opposite charges. We will have
much more to say about the applications of dipoles in Chapter 10.
Turning to mathematics, we introduce the divergence, which
gives a measure of the flux of a vector field out of a small volume.
We prove Gauss’s theorem (or the divergence theorem) and then
use it to write Gauss’s law in differential form. The result is the first
of the four equations known as Maxwell’s equations (the subject
of Chapter 9). We explicitly calculate the divergence in Cartesian
coordinates. The divergence of the gradient is known as the Lapla-
cian operator. Functions whose Laplacian equals zero have many
important properties, one of which leads to Earnshaw’s theorem,
which states that it is impossible to construct a stable electrostatic
equilibrium in empty space. We introduce the curl, which gives a
measure of the line integral of a vector field around a small closed
curve. We prove Stokes’ theorem and explicitly calculate the curl in
Cartesian coordinates. The conservative nature of a static electric
2.1 Line integral of the electric field
59
field implies that its curl is zero. See Appendix F for a discussion
of the various vector operators in different coordinate systems.
2.1 Line integral of the electric field
Suppose that E is the field of some stationary distribution of electric
charges. Let P\ and Pi denote two points anywhere in the field. The line
integral of E between the two points is fp j 2 E • c/s, taken along some path
that runs from P\ to Pi, as shown in Fig. 2.1. This means: divide the
chosen path into short segments, each segment being represented by a
vector connecting its ends; take the scalar product of the path-segment
vector with the field E at that place; add these products up for the whole
path. The integral as usual is to be regarded as the limit of this sum as
the segments are made shorter and more numerous without limit.
Let’s consider the field of a point charge q and some paths running
from point Pi to point Pi in that field. Two different paths are shown in
Fig. 2.2. It is easy to compute the line integral of E along path A, which
is made up of a radial segment running outward from Pi and an arc of
Figure 2.1.
Showing the division of the path into path
elements els.
Figure 2.2.
The electric field E is that of a positive point
charge q. The line integral of E from Pi to Pi
along path A has the value {q/4neo){\/r\ - 1 ~/r 2 ).
It will have exactly the same value if calculated
for path B, or for any other path from P { to P 2 .
60
The electric potential
radius r^. Along the radial segment of path A, E and ds are parallel, the
magnitude of E is q/Aite^r 2 , and E • ds is simply (q / Att eor 2 ) ds. Thus
the line integral on that segment is
( 2 . 1 )
The second leg of path A, the circular segment, gives zero because E is
perpendicular to ds everywhere on that arc. The entire line integral is
therefore
( 2 . 2 )
Now look at path B. Because E is radial with magnitude q/4it€or 2 ,
E ■ ds — ( q/4iT€or 2 ) dr even when ds is not radially oriented. The corres¬
ponding pieces of path A and path B indicated in the diagram make iden¬
tical contributions to the integral. The part of path B that loops beyond
n_ makes a net contribution of zero; contributions from corresponding
outgoing and incoming parts cancel. For the entire line integral, path B
will give the same result as path A. As there is nothing special about path
B, Eq. (2.1) must hold for any path running from P\ to Pn.
Here we have essentially repeated, in different language, the argu¬
ment in Section 1.5, illustrated in Fig. 1.5, concerning the work done in
moving one point charge near another. But now we are interested in the
total electric field produced by any distribution of charges. One more
step will bring us to an important conclusion. The line integral of the
sum of fields equals the sum of the line integrals of the fields calculated
separately. Or, stated more carefully, if E = Ei + E 2 + • • •, then
rPi rPi r p 2
/ E • ds = I Ei-d/s + / E 2 • ds + • • • , (2.3)
JPi JP\ JPi
where the same path is used for all the integrations. Now any electro¬
static field can be regarded as the sum of a number (possibly enormous)
of point-charge fields, as expressed in Eq. (1.20) or Eq. (1.22). There¬
fore if the line integral from P\ to Pj is independent of path for each
of the point-charge fields Ej,E 2 , ..., the total field E must have this
property:
The line integral fp j 2 E ■ ds for any given electrostatic field E has
the same value for all paths from Pi to P 2 .
2.2 Potential difference and the potential function
61
The points P 2 and P\ may coincide. In that case the paths are all
closed curves, among them paths of vanishing length. This leads to the
following corollary:
The line integral / E ■ ds around any closed path in an electrostatic
field is zero.
By electrostatic field we mean, strictly speaking, the electric field of
stationary charges. Later on, we shall encounter electric fields in which
the line integral is not path-independent. Those fields will usually be
associated with rapidly moving charges. For our present purposes we
can say that, if the source charges are moving slowly enough, the field
E will be such that / E ■ ds is practically path-independent. Of course,
if E itself is varying in time, the E in f E • ds must be understood as
the field that exists over the whole path at a given instant of time. With
that understanding we can talk meaningfully about the line integral in a
changing electrostatic field.
2.2 Potential difference and the potential function
Because the line integral in the electrostatic field is path-independent,
we can use it to define a scalar quantity (p 2 i, without specifying any par¬
ticular path:
rPi
<p 2 l — — I E ■ d/s.
JPI
(2.4)
With the minus sign included here, <p 21 is the work per unit charge done
by an external agency in moving a positive charge from Pi to P 2 in the
field E. (The external agency must supply a force F ex t = — gE to balance
the electrical force F e i e c = <?E; hence the minus sign.) Thus (j)i\ is a
single-valued scalar function of the two positions Pi and P 2 . We call it
the electric potential difference between the two points.
In our SI system of units, potential difference is measured in joule/
coulomb. This unit has a name of its own, the volt:
1 volt = 1
joule
coulomb
(2.5)
One joule of work is required to move a charge of one coulomb through a
potential difference of one volt. In the Gaussian system of units, potential
difference is measured in erg/esu. This unit also has a name of its own,
the statvolt (“stat” comes from “electrostatic”). As an exercise, you can
use the 1 C ^ 3 • 10 9 esu relation from Section 1.4 to show that one volt
is equivalent to approximately 1/300 statvolt. These two relations are
accurate to better than 0.1 percent, thanks to the accident that c is that
62
The electric potential
close to 3 ■ 10 s m/s. Appendix C derives the conversion factors between
all of the corresponding units in the SI and Gaussian systems. Further
discussion of the exact relations between SI and Gaussian electrical units
is given in Appendix E, which takes into account the definition of the
meter in terms of the speed of light.
Suppose we hold Pi fixed at some reference position. Then 0 21
becomes a function of Pi_ only, that is, a function of the spatial coord¬
inates x, y, z. We can write it simply 0 (x, y, z), without the subscript,
if we remember that its definition still involves agreement on a refer¬
ence point Pi. We can say that r/j is the potential associated with the
vector field E. It is a scalar function of position, or a scalar field (they
mean the same thing). Its value at a point is simply a number (in units of
work per unit charge) and has no direction associated with it. Once the
vector field E is given, the potential function 0 is determined, except
for an arbitrary additive constant allowed by the arbitrariness in our
choice of Pi.
Example Find the potential associated with the electric field described in
Fig. 2.3, the components of which are E x — Ky, E y = Kx, E z = 0, with K a
constant. This is a possible electrostatic field; we will see why in Section 2.17.
Some field lines are shown.
Solution Since E z = 0, the potential will be independent of z and we need
consider only the xy plane. Let xi, y\ be the coordinates of Pi, and x 2 , y 2 the
coordinates of P 2 . It is convenient to locate Pi at the origin: xi = 0, yi = 0.
To evaluate — /E • ds from this reference point to a general point (X2,y2) it is
easiest to use a path like the dashed path ABC in Fig. 2.3:
Figure 2.3.
A particular path, ABC , in the electric field
E x = Ky, E y = Kx. Some field lines are shown.
<P(x 2 A 2 ) = -
r(x2,yi)
r(x 2 , 0 )
r(x 2 ,yi)
I
II
c n
w
/ E x dx -
/
2(0,0}
J (0,0)
J(X 2 , 0)
Eydy. (2.6)
The first of the two integrals on the right is zero because E x is zero along the x
axis. The second integration is carried out at constant x, with Ey = Kx 2 \
r(jc 2 ,yi) ry 2
/ Eydy = - l
J(xo. 0) Jo
'(*2,0)
Kx 2 dy = -Kx 2 y 2 .
(2.7)
There was nothing special about the point (x 2 ,y 2 ) so we can drop the subscripts:
</>(x,y) = -Kxy (2.8)
for any point (x,y) in this field, with zero potential at the origin. Any constant
could be added to this. That would only mean that the reference point to which
zero potential is assigned had been located somewhere else.
Example (Potential due to a uniform sphere) A sphere has radius R and
uniform volume charge density p. Use the results from the example in Section 1.11
to find the potential for all values of r. both inside and outside the sphere. Take
the reference point Pj to be infinitely far away.
2.3 Gradient of a scalar function
63
Solution From the example in Section 1.11, the magnitude of the (radial) elec¬
tric field inside the sphere is E(r ) = pr/3eo , and the magnitude outside is
E(r) — pR? /3e(p' 2 - Equation (2.4) tells us that the potential equals the negative
of the line integral of the field, from P\ (which we are taking to be at infinity)
down to a given radius r. The potential outside the sphere is therefore
C r r 0 r3
<Pout(r) = - E{r') dr' = - - - dr' =
Joo Joo 3e 0 r ,z
pR_
3 e 0 r'
(2.9)
In terms of the total charge in the sphere, Q = (4nR?/3)p, this potential is sim¬
ply <t>out(r) = Q/Ane^r. This is as expected, because we already knew that the
potential energy of a charge q due to the sphere is qQ/4jreQr. And the potential
(f> equals the potential energy per unit charge.
To find the potential inside the sphere, we must break the integral into two
pieces:
0in W = - ( R E(r') dr' - f E(r') dr' = - f dr' - f BL. d r'
Joo Jr Joo 3e 0 E z Jr 3e 0
P R3 P (r 2 j^2\ P R2 P r2
3 (qR 6 ep 2e 0 6 eo
( 2 . 10 )
Note that Eqs. (2.9) and (2.10) yield the same value of <J> at the surface of the
sphere, namely <p(R) — pR~/ 3eo- So (f> is continuous across the surface, as it
should be. (The field is everywhere finite, so the line integral over an infinitesimal
interval must yield an infinitesimal result.) The slope of 0 is also continuous,
because E(r) (which is the negative derivative of <p, because (j) is the negative
integral of E) is continuous. A plot of <f>(r) is shown in Fig. 2.4.
The potential at the center of the sphere is (p( 0) = pR 2 /2eo, which is 3/2
times the value at the surface. So if you bring a charge in from infinity, it takes
2/3 of your work to reach the surface, and then 1 /3 to go the extra distance of R
to the center.
0 (r)
Figure 2.4.
The potential due to a uniform sphere of charge.
We must be careful not to confuse the potential (p associated with a
given field E with the potential energy of a system of charges. The poten¬
tial energy of a system of charges is the total work required to assemble
it, starting with all the charges far apart. In Eq. (1.14), for example, we
expressed U, the potential energy of the charge system in Fig. 1.6. The
electric potential <p(x,y,z) associated with the field in Fig. 1.6 would
be the work per unit charge required to move a unit positive test charge
from some chosen reference point to the point (x, y, z ) in the field of that
structure of nine charges.
2.3 Gradient of a scalar function
Given the electric field, we can find the electric potential function. But
we can also proceed in the other direction; from the potential we can
derive the field. It appears from Eq. (2.4) that the field is in some sense
the derivative of the potential function. To make this idea precise we
introduce the gradient of a scalar function of position. Fet f(x, y, z) be
64
The electric potential
(a)
Figure 2.5.
The scalar function f(x,y) is represented by the
surface in (a). The arrows in (b) represent the
vector function, grad/.
some continuous, differentiable function of the coordinates. With its par¬
tial derivatives df/dx, df/dy, and df/dz we can construct at every point
in space a vector, the vector whose x, y, z components are equal to the
respective partial derivatives. 1 This vector we call the gradient of/, writ¬
ten “grad/,” or V/:
w ~df df df
v/ = x— +y— +z—.
ox cry oz
(2.13)
V/ is a vector that tells how the function / varies in the neighborhood
of a point. Its x component is the partial derivative of / with respect to
x, a measure of the rate of change off as we move in the x direction.
The direction of the vector V/ at any point is the direction in which one
must move from that point to find the most rapid increase in the function
/. Suppose we were dealing with a function of two variables only, x and
y, so that the function could be represented by a surface in three dimen¬
sions. Standing on that surface at some point, we see the surface rising
in some direction, sloping downward in the opposite direction. There is a
direction in which a short step will take us higher than a step of the same
length in any other direction. The gradient of the function is a vector in
that direction of steepest ascent, and its magnitude is the slope measured
in that direction.
Figure 2.5 may help you to visualize this. Suppose some particular
function of two coordinates x and y is represented by the surface//*;, y)
sketched in Fig. 2.5(a). At the location (xi,yi) the surface rises most
steeply in a direction that makes an angle of about 80° with the positive
x direction. The gradient of f(x,y), V/, is a vector function of x and y.
Its character is suggested in Fig. 2.5(b) by a number of vectors at various
points in the two-dimensional space, including the point (x i, y i) • The
vector function V/ defined in Eq. (2.13) is simply an extension of this
idea to three-dimensional space. (Be careful not to confuse Fig. 2.5(a)
with real three-dimensional xyz space; the third coordinate there is the
value of the function f(x,y).)
As one example of a function in three-dimensional space, suppose/
is a function of r only, where r is the distance from some fixed point O.
On a sphere of radius ro centered about O, f = /(ro) is constant. On a
slightly larger sphere of radius ro + dr it is also constant, with the value
/ =/(ro + dr). If we want to make the change from/(ro) to/(ro + dr),
1 We remind the reader that a partial derivative with respect to x, of a function of x, y, z,
written simply df/dx, means the rate of change of the function with respect to x with
the other variables y and z held constant. More precisely,
df __ |jm f(x+Ax, y,z) — f(x,y, z)
Hx Ax->0 A.t
As an example, if / = x/y/,
( 2 . 11 )
2.5 Potential of a charge distribution
65
the shortest step we can make is to go radially (as from A to B ) rather
than from A to C, in Fig. 2.6. The “slope” of / is thus greatest in the
radial direction, so V/ at any point is a radially pointing vector. In fact
V/ = r (df /dr) in this case, f denoting, for any point, a unit vector in the
radial direction. See Section F.2 in Appendix F for further discussion of
the gradient.
2.4 Derivation of the field from the potential
It is now easy to see that the relation of the scalar function/ to the vector
function V/ is the same, except for a minus sign, as the relation of the
potential / to the field E. Consider the value of <j) at two nearby points,
(x, y, z) and (x + dx,y + dy , z + dz) . The change in /, going from the first
point to the second, is, in first-order approximation,
3 eh 3 d> 3 (h
d$ = — dx + — dy + — dz.
dx ay dz
(2.14)
On the other hand, from the definition of </> in Eq. (2.4), the change can
also be expressed as
B
Figure 2.6.
The shortest step for a given change in/ is the
radial step AB, if/ is a function of r only.
d<j> = — E ■ ds.
(2.15)
The infinitesimal vector displacement ds is just x dx + y dy + z dz. Thus
if we identify E with — V/, where V/ is defined via Eq. (2.13), then
Eqs. (2.14) and (2.15) become identical. So the electric field is the nega¬
tive of the gradient of the potential:
(2.16)
The minus sign came in because the electric field points from a region of
greater potential toward a region of lesser potential, whereas the vector
V/ is defined so that it points in the direction of increasing <p.
To show how this works, we go back to the example of the field
in Fig. 2.3. From the potential given by Eq. (2.8), <p = —Kxy, we can
recover the electric field we started with:
E = -V(-foy) = -
(~Kxy) =K(xy + yx).
(2.17)
2.5 Potential of a charge distribution
We already know the potential that goes with a single point charge,
because we calculated the work required to bring one charge into the
neighborhood of another in Eq. (1.9). The potential at any point, in the
field of an isolated point charge q , is just q/Aixe^r, where r is the distance
66
The electric potential
Figure 2.7.
Each element of the charge distribution
p<y,/ ,z!) contributes to the potential <j> at the
point (x,y,z ). The potential at this point is the
sum of all such contributions; see Eq. (2.18).
from the point in question to the source q, and where we have assigned
zero potential to points infinitely far from the source.
Superposition must work for potentials as well as fields. If we have
several sources, the potential function is simply the sum of the poten¬
tial functions that we would have for each of the sources present alone -
providing we make a consistent assignment of the zero of potential in
each case. If all the sources are contained in some finite region, it is
always possible, and usually the simplest choice, to put zero potential at
infinite distance. If we adopt this rule, the potential of any charge distri¬
bution can be specified by the integral
(p(x,y, z)
f p (x 7 , y, z!) dx' dy' dz'
J all 47T€or
(2.18)
where r is the distance from the volume element dx dy dz' to the point
(x,y,z) at which the potential is being evaluated (Fig. 2.7). That is, r =
[(x — x 7 ) 2 + (y — y') 1 + (z — z 7 ) 2 ] 1 / 2 . Notice the difference between
this and the integral giving the electric field of a charge distribution; see
Eq. (1.22). Here we have r in the denominator, not r 2 , and the integral
is a scalar not a vector. From the scalar potential function <p(x,y,z) we
can always find the electric field by taking the negative gradient of (f >,
according to Eq. (2.16).
In the case of a discrete distribution of source charges, the above
integral is replaced by a sum over all the charges, indexed by i:
4> (x, y, z)
E
all sources
qt
(2.19)
where r is the distance from the charge q ,■ to the point (x,y, z).
Example (Potential Of two point charges) Consider a very simple exam¬
ple, the potential of the two point charges shown in Fig. 2.8. A positive charge of
12 pC is located 3 m away from a negative charge, —6 pC. (The “p” prefix stands
for “micro,” or 10 6 .) The potential at any point in space is the sum of the poten¬
tials due to each charge alone. The potentials for some selected points in space
are given in the diagram. No vector addition is involved here, only the algebraic
addition of scalar quantities. For instance, at the point on the far right, which is
6 m from the positive charge and 5 m from the negative charge, the potential has
the value
1 /l2-10 _6 C -6- 10“ 6 C\ 0.8- 10“ 6 C/m
4tt€q V 6 m 5 m I 4jreo
= 7.2 ■ 10 3 J/C = 7.2 • 10 3 V, (2.20)
where we have used l/47reo sa9 ■ 10 9 Nm 2 /C 2 (and also lNm=lJ). The
potential approaches zero at infinite distance. It would take 7.2 • 10 3 J of work
2.5 Potential of a charge distribution
67
Figure 2.8.
The electric potential <t> at various points in a
system of two point charges. f goes to zero at
infinite distance and is given in units of volts, or
joules per coulomb.
to bring a unit positive charge in from infinity to a point where <j> = 7.2 • 10 3 V.
Note that two of the points shown on the diagram have <j> = 0. The net work
done in bringing in any charge to one of these points would be zero. You can see
that there must be an infinite number of such points, forming a surface in space
surrounding the negative charge. In fact, the locus of points with any particular
value of <f> is a surface - an equipotential surface - which would show on our
two-dimensional diagram as a curve.
There is one restriction on the use of Eq. (2.18): it may not work
unless all sources are confined to some finite region of space. A simple
example of the difficulty that arises with charges distributed out to infi¬
nite distance is found in the long charged wire whose field E we studied
in Section 1.12. If we attempt to carry out the integration over the charge
distribution indicated in Eq. (2.18), we find that the integral diverges -
we get an infinite result. No such difficulty arose in finding the electric
field of the infinitely long wire, because the contributions of elements of
the line charge to the field decrease so rapidly with distance. Evidently
we had better locate the zero of potential somewhere close to home, in
a system that has charges distributed out to infinity. Then it is simply
a matter of calculating the difference in potential , between the gen¬
eral point (x, y, z) and the selected reference point, using the fundamental
relation, Eq. (2.4).
Example (Potential of a long charged wire) To see how this goes in the
case of the infinitely long charged wire, let us arbitrarily locate the reference
point P i at a distance r\ from the wire. Then to carry a charge from P\ to
68
The electric potential
any other point Pi at distance rj requires the work per unit charge, using
Eq. (1.39):
rPi rn ( \ \
021 = - / Eds = - - - dr
JPi Jri \2ne 0 rJ
= ---— lnr 2 + Inn. (2-21)
2jteo 2 ti cq
This shows that the electric potential for the charged wire can be taken as
X
0 =-In r + constant. (2.22)
2jre 0
The constant, (X/2neg) In r\ in this case, has no effect when we take —grad0 to
get back to the field E. In this case,
„ . d<b Xr
E = -V0 = -r— = -.
dr 2neg r
(2.23)
2.6 Uniformly charged disk
Let us now study the electric potential and field around a uniformly
charged disk. This is a charge distribution like that discussed in
Section 1.13, except that it has a limited extent. The flat disk of radius
a in Fig. 2.9 carries a positive charge spread over its surface with the
constant density er, in C/m 2 . (This is a single sheet of charge of infinites¬
imal thickness, not two layers of charge, one on each side. That is, the
total charge in the system is jta 2 o.) We shall often meet surface charge
distributions in the future, especially on metallic conductors. However,
the object just described is not a conductor; if it were, as we shall soon
see, the charge could not remain uniformly distributed but would redis¬
tribute itself, crowding more toward the rim of the disk. What we have
is an insulating disk, like a sheet of plastic, upon which charge has been
“sprayed” so that every square meter of the disk has received, and holds
fixed, the same amount of charge.
Example (Potential on the axis) Let us find the potential due to our uni¬
formly charged disk, at some point P\ on the axis of symmetry, which we have
made the y axis. All charge elements in a thin, ring-shaped segment of the disk
lie at the same distance from P\. If s denotes the radius of such an annular seg¬
ment and ds is its width, its area is 2ns ds. The amount of charge it contains, dq,
is therefore dq = a 2ns ds. Since all parts of this ring are the same distance away
from P i, namely, r = yjy 2 + s 2 , the contribution of the ring to the potential at
P\ is dq/Anegr = osds/ (2eg \/y 2 + s 2 ). To get the potential due to the whole
disk, we have to integrate over all such rings:
0(0, y,0)
/
dq
Anegr
osds
2e 0 V}’ 2 + s 2
a
2<D
Jy 2 + s 2
Figure 2.9.
Finding the potential at a point P\ on the axis of
a uniformly charged disk.
o
(2.24)
2.6 Uniformly charged disk
69
Putting in the limits, we obtain
0(0, v, 0) = -— y 2 + a 2 — y^ fory > 0. (2.25)
A minor point deserves a comment. The result we have written down in
Eq. (2.25) holds for all points on the positive y axis. It is obvious from the phys¬
ical symmetry of the system (there is no difference between one face of the disk
and the other) that the potential must have the same value for negative and pos¬
itive y, and this is reflected in Eq. (2.24), where only y 2 appears. But in writing
Eq. (2.25) we made a choice of sign in taking the square root of y 2 , with the
consequence that it holds only for positive y. The correct expression for y < 0 is
obtained by the other choice of root and is given by
0(0,y,0) = {^Jy 2 + a- + y^ fory < 0. (2.26)
In view of this, we should not be surprised to find a kink in the plot of 0(0,y, 0)
at y = 0. Indeed, the function has an abrupt change of slope there, as we see in
Fig. 2.10, where we have plotted as a function of y the potential on the axis. The
potential at the center of the disk is
0(0,0,0)=—. (2.27)
2eo
This much work would be required to bring a unit positive charge in from infinity,
by any route, and leave it sitting at the center of the disk.
The behavior of 0(0,y,0) for very large y is interesting. For y>a we can
approximate Eq. (2.25) as follows:
y 2 + a 2 — y = y
(2.28)
70
The electric potential
Hence
4>(0,y,0) for )'»a. (2.29)
4e 0 y
Now 7 za 2 a is the total charge q on the disk, and Eq. (2.29), which can be written
as 7ta 2 a/4neoy, is just the expression for the potential due to a point charge of
this magnitude. As we should expect, at a considerable distance from the disk
(relative to its diameter), it doesn’t matter much how the charge is shaped; only
the total charge matters, in first approximation. In Fig. 2.10 we have drawn, as a
dashed curve, the function a 2 a/4egy. You can see that the axial potential func¬
tion approaches its asymptotic form pretty quickly.
It is not quite so easy to derive the potential for general points away
from the axis of symmetry, because the definite integral isn’t so simple.
It proves to be something called an elliptic integral. These functions are
well known and tabulated, but there is no point in pursuing here mathe¬
matical details peculiar to a special problem. However, one further cal¬
culation, which is easy enough, may be instructive.
Example (Potential on the rim) We can find the potential at a point on the
very edge of the disk, such as Pi in Fig. 2.11. To calculate the potential at Pi we
can consider first the thin wedge of length R and angular width d9, as shown.
An element of the wedge, the black patch at distance r from P 2 , contains an
amount of charge dq = ard9 dr. Its contribution to the potential at Pi is there¬
fore dq/4neo r = a d9 dr/4neQ. The contribution of the entire wedge is then
r R
(a d9/4neQ ) / dr = (aR/4neo) d9. Now R is 2acos9, from the geometry of
JO
the right triangle, and the whole disk is swept out as 9 ranges from —tc/2 to 7r/2.
Thus we find the potential at P 2 :
aa /’ 7r / 2 aa
cp = - / cos 9d9 = -. (2.30)
2ne 0 J—ji/ 2 xeo
Comparing this with the potential at the center of the disk, aa/2 cq, we see
that, as we should expect, the potential falls off from the center to the edge of the
disk. The electric field, therefore, must have an outward component in the plane
of the disk. That is why we remarked earlier that the charge, if free to move,
would redistribute itself toward the rim. To put it another way, our uniformly
charged disk is not a surface of constant potential, which any conducting surface
must be unless charge is moving.-
Figure 2.11.
Finding the potential at a point P 2 on the rim of a 2 ^ r ... ... , , . ,
, 1 i- 1 The tact that conducting surfaces have to be equipotentials will be discussed
uniformly Charged disk. thoroughly in Chapter 3
2.6 Uniformly charged disk
71
Let us now examine the electric field due to the disk. For y > 0, the
field on the symmetry axis can be computed directly from the potential
function given in Eq. (2.25):
cr
y > 0. (2.31)
2eo
To be sure, it is not hard to compute E y directly from the charge distri¬
bution, for points on the axis. We can again slice the disk into concentric
rings, as we did prior to Eq. (2.24). But we must remember that E is
a vector and that only the y component survives in the present setup,
whereas we did not need to worry about components when calculating
the scalar function </> above.
As y approaches zero from the positive side, E y approaches er/2eo.
On the negative y side of the disk, which we shall call the back, E points
in the other direction and its y component E y is —a/2eo- This is the
same as the field of an infinite sheet of charge of density a, derived in
Section 1.13. It ought to be, for at points close to the center of the disk,
the presence or absence of charge out beyond the rim can’t make much
difference. In other words, any sheet looks infinite if viewed from close
up. Indeed, E y has the value a/ 2eo not only at the center, but also all
over the disk.
For large y, we can find an approximate expression for E y by using
a Taylor series approximation as we did in Eq. (2.28). You can show that
E y approaches a 2 a/4-€oy 2 , which can be written as tt a 2 cr/A tz e^y 1 . This
is correctly the field due to a point charge with magnitude jta 2 a.
In Fig. 2.12 we show some field lines for this system and also, plotted
as dashed curves, the intersections on the yz plane of the surfaces of
constant potential. Near the center of the disk these are lens-like surfaces,
while at distances much greater than a they approach the spherical form
of equipotential surfaces around a point charge.
Figure 2.12 illustrates a general property of field lines and equipoten¬
tial surfaces. A field line through any point and the equipotential surface
through that point are perpendicular to one another, just as, on a con¬
tour map of hilly terrain, the slope is steepest at right angles to a contour
of constant elevation. This must be so, because if the field at any point
had a component parallel to the equipotential surface through that point,
it would require work to move a test charge along a constant-potential
surface.
The energy associated with this electric field could be expressed as
the integral over all space of (eo/2 )E 2 dv. It is equal to the work done in
assembling this distribution, starting with infinitesimal charges far apart.
In this particular example, as Exercise 2.56 will demonstrate, that work
72
The electric potential
Figure 2.12.
The electric field of the uniformly charged disk.
Solid curves are field lines. Dashed curves are
intersections, with the plane of the figure, of
surfaces of constant potential.
is not hard to calculate directly if we know the potential at the rim of a
uniformly charged disk.
There is a general relation between the work U required to assem¬
ble a charge distribution p(x,y, z) and the potential (p(x,y,z ) of that
distribution:
p4> dv
(2.32)
Equation (1.15), which gives the energy of a system of discrete point
charges, could have been written in this way:
1 N
qk
2 “ “ 4jre 0 rjk
(2.33)
The second sum is the potential at the location of the jth charge, due to all
the other charges. To adapt this to a continuous distribution we merely
2.7 Dipoles
73
replace qj with p dv and the sum over j by an integral, thus obtaining
Eq. (2.32).
2.7 Dipoles
Consider a setup with two equal and opposite charges ±q located at
positions ±1 /2 on the y axis, as shown in Fig. 2.13. This configura¬
tion is called a dipole. The purpose of this section is to introduce the
basics of dipoles. We save further discussion for Chapter 10, where we
define the word “dipole” more precisely, derive things in more general¬
ity, and discuss examples of dipoles in actual matter. For now we just
concentrate on determining the electric field and potential of a dipole.
We have all of the necessary machinery at our disposal, so let’s see what
we can find.
We will restrict the treatment to points far away from the dipole
(that is, points with r 'P> i). Although it is easy enough to write down an
exact expression for the potential cp (and hence the field E = — Vr/>) at
any position, the result isn’t very enlightening. But when we work in the
approximation of large distances, we obtain a result that, although isn’t
exactly correct, is in fact quite enlightening. That’s how approximations
work - you trade a little bit of precision for a large amount of clarity.
Our strategy will be to find the potential (p in polar (actually spheri¬
cal) coordinates, and then take the gradient to find the electric field E. We
then determine the shape of the field-line and constant-potential curves.
To make things look a little cleaner in the calculations below, we write
1 /4jt€o as k in some intermediate steps.
2.7.1 Calculation of 0 and E
First note that, since the dipole setup is rotationally symmetric around
the line containing the two charges, it suffices to find the potential in an
arbitrary plane containing this line. We will use spherical coordinates,
which reduce to polar coordinates in a plane because the angle (p doesn’t
come into play (but note that 9 is measured down from the vertical axis).
Consider a point P with coordinates (r, 0), as shown in Fig. 2.14. Let
r i and n be the distances from P to the two charges. Then the exact
expression for the potential at P is (with k = 1 /4-jteo)
<p P =
kq
n
kq
n
(2.34)
If desired, the law of cosines can be used to write r\ and n in terms of r,
6, and l.
Let us now derive an approximate form of this result, valid in the
r i limit. One way to do this is to use the law-of-cosines expressions
for r\ and rp, this is the route we will take in Chapter 10. But for the
present purposes a simpler method suffices. In the r ~S> l limit, a closeup
view of the dipole is shown in Fig. 2.15. The two lines from the charges
to P are essentially parallel, so we see from the figure that the lengths of
y
O?
in
m
Figure 2.13.
Two equal and opposite charges form a dipole.
p
Figure 2.14.
Finding the potential <p at point P.
Figure 2.15.
Closeup view of Fig. 2.14.
74
The electric potential
-q q
9-•
•- 6
Figure 2.16.
Two possible kinds of quadrupoles.
these lines are essentially r\ = r — {1/ 2) cos 9 and rj = r+ {1/2) cos 9.
Using the approximation 1/(1 ± e) « 1 e, Eq. (2.34) becomes
4>{r,9) =
1
l cos 9
2 r -I
(2.35)
where p = ql is called the dipole moment.
There are three important things to note about this result. First,
<p(r,9) depends on q and l only through their product, p = ql. This
means that if we make q ten times larger and l ten times smaller, the
potential at a given point P stays the same (at least in the r ~S> l approx¬
imation). An idealized dipole or point dipole is one where l —>■ 0 and
q -> oo, with the product p — ql taking on a particular finite value. In
the other extreme, if we make q smaller and l proportionally larger, the
potential at P again stays the same. Of course, if we make i too large,
our r^> l assumption eventually breaks down.
Second, <p{r,9) is proportional to 1/r 2 , in contrast with the 1 / r
dependence for a point-charge potential. We will see below that the
present 1/r 2 dependence in <p{r, 9 ) leads to an E field that falls off like
1/r 3 , in contrast with the 1/r 2 dependence for a point-charge field. It
makes sense that the potential (and field) falls off faster for a dipole,
because the potentials from the two opposite point charges nearly cancel.
The dipole potential is somewhat like the derivative of the point-charge
potential, in that we are taking the difference of two nearby values.
Third, there is angular dependence in <p{r,9), in contrast with the
point-charge potential. This is expected, in view of the fact that the dipole
has a preferred direction along the line joining the charges, whereas a
point charge has no preferred direction.
We will see in Chapter 10 that the q/r point-charge (or monopole)
potential and the ql/r 2 dipole potential (just looking at the r dependence)
are the first two pieces of what is called the multipole expansion. A gen¬
eral charge distribution also has a quadrupole term in the potential that
goes like ql 2 /r 3 (where l is some length scale of the system), and an
octupole term that goes like ql? /r 4 , and so on. These pieces have more
complicated angular dependences. Two examples of quadrupole arrange¬
ments are shown in Fig. 2.16. A quadrupole is formed by placing two
oppositely charged dipoles near each other, just as a dipole is formed by
placing two oppositely charged monopoles near each other. The various
terms in the expansion are called the moments of the distribution.
2.7 Dipoles
75
Even the simple system of the dipole shown in Fig. 2.13 has higher
terms in its multipole expansion. If you keep additional terms in the
1/(1 ± e) Taylor series in Eq. (2.35), you will find that the quadrupole
term is zero, but the octupole term is nonzero. It is easy to see that the
terms with even powers of r are nonzero. However, in the limit of an
idealized dipole (£ —* 0 and q -> oo, with ql fixed), only the dipole
potential survives, because the higher-order terms are suppressed by
additional powers of l/r.
Along the same lines, we can back up a step in the expansion and
consider the monopole term. If an object has a nonzero net charge (note
that our dipole does not), then the monopole potential, q/r , dominates,
and all higher-order terms are comparatively negligible in the r l limit.
The distribution of charge in an object determines which of the terms in
the expansion is the first nonzero one, and it is this term that determines
the potential (and hence field) at large distances. We label the object
according to the first nonzero term; see Fig. 2.17.
Let’s now find the electric field, E = — V0, associated with the
dipole potential in Eq. (2.35). In spherical coordinates (which reduce
to polar coordinates in this discussion) the gradient of </> is V</> =
r(3</>/3r) + f?(l/r)(3</>/30); see Appendix F. So the electric field is
E (r,9)
„ 3 / kql cos 9 \ ~ 1 3 / kq£ cos 9
dr \ r 2 ) r d9 \ r 2
—^ (2 cos 9 r + sin 9 0)
— -^ , (2 cos 9 r + sin 6 0)
47re 0 r 3 v ’
= —-— 7 (2 cos 9 r + sinff 6).
47T€ 0 r 3 v ’
(2.36)
Monopole
t c i
O ~q
Dipole
Quadrupole
Figure 2.17.
Examples of different objects in the multipole
expansion.
A few field lines are shown in Fig. 2.18. Let’s look at some special
cases for 9. Equation (2.36) says that E points in the positive radial direc¬
tion for 9 = 0 and the negative radial direction for 9 = it. These facts
imply that E points upward everywhere on the y axis. Equation (2.36)
also says that E points in the positive tangential direction for 9 = n/2
and the negative tangential direction for 9 = 3ic /2. In view of the local f
and 0 basis vectors shown in Fig. 2.18 (which vary with position, unlike
the Cartesian x and y basis vectors), this means that E points down¬
ward everywhere on the x axis. We haven’t drawn the lines for small r,
to emphasize the fact that our results are valid only in the limit r i.
There is a field for small r, of course (and it diverges near each charge);
it’s just that it doesn’t take the form given in Eq. (2.36).
76
The electric potential
Figure 2.18.
Electric field lines for a dipole. Note that the r
and 6 basis vectors depend on position.
Figure 2.19.
Field lines and constant-potential curves for a
dipole. The two sets of curves are orthogonal at
all intersections. The solid l ines show
constant-^ curves (r = r o Vcos0), and the
dashed lines show E field lines (r = ? 0 sin 2 9).
2 . 7.2 The shapes of the curves
Let us now be quantitative about the shape of the E and 0 curves. More
precisely, let us determine the equations that describe the field-line curves
and the constant-potential curves. In the process we will also determine
the slopes of the tangents to these curves. We know that the two classes
of curves are orthogonal wherever they meet, because E is the (negative)
gradient of 0, and because the gradient of a function is always perpendicular
to the level-surface curves. This orthogonality is evident in Fig. 2.19. Our
task now is to derive the two expressions for r given in this figure.
Let’s look at 0 first. We will find the equation for the constant-
potential curves and then use this to find the slope of the tangent at
any point. The equation for the curves is immediately obtained from
Eq. (2.35). The set of points for which 0 takes on the constant value
00 is given by
kq£ cos 6
= 00
(kql\ .
- cos 0 =>-
r — ro\/ cos 9
Wo /
(2.37)
where ro = ^Jkqt /0o is the radius associated with the angle 9 — 0. This
result is valid in the upper half-plane where —Tt/2<9<n/2. In the
lower half-plane, both 0o and cos 9 are negative, so we need to add in some
absolute-value signs. That is, r — roVI cos 9\, where ro = ^ /kql /10o|■
The constant-potential curves in Fig. 2.19 are the intersections of the
constant-potential surfaces with the plane of the paper. These surfaces
are generated by rotating the curves around the vertical axis. The curves
are stretched horizontally compared with the circles described by the
relation r — ro cos 9 (which you can verify is indeed a circle).
12.10 Chapter 10
745
We see that t — RC is independent of A , because R oc 1/ A and C a
A. (Basically, if a given patch of the membrane leaks its charge on
a given time scale, then putting a bunch of these patches together
shouldn’t change the time scale, because each patch doesn’t care that
there are others next to it.) Using the information given for 1 cm 2 of
the membrane, we have t = RC = (1000 £2)(10 -6 F) = 10 -3 s.
Since R = ps/A, the resistivity is given by
P =
RA
s
(1000 fi)(10 -4 m 2 )
2.7 • 10 -9 m
4 ■ 10 7 ohm-m.
(12.453)
From Fig. 4.8, this is a little more than 100 times the resistivity of
pure water.
10.2 Force on a dielectric
(a) The equivalent capacitance of two capacitors in parallel is simply the
sum of the capacitances. (The rule is opposite to that for resistors;
see Problem 3.18.) The capacitance of the part with the dielectric is
k times what it would be if there were vacuum there. So the total
capacitance is given by
c = c 1 + c 2 = ^ + ^
S S
= 6ofl (fe-*) + ^ = ! o«
s s s L J
The stored energy is then
(12.454)
Q 2 s
2 (Qa[b + (k — l)x]
(12.455)
Note that as x changes, the charge stays constant (by assumption), but
the potential does not. So the Q<j>/2 and C<p~/2 forms of the energy
aren’t useful.
(b) The force is
dU _ Q 2 s(k - 1 )
dx 2 cq a[b + (at — l).r] 2
(12.456)
The positive sign here means that the force points in the direction of
increasing x. That is, the dielectric slab is pulled into the capacitor.
But it’s risky to trust this sign blindly. Physically, the force points in
the direction of decreasing energy. And we see from the above expres¬
sion for U that the energy decreases as x increases (because k > 1).
The force F is correctly zero if k = 1, because in that case we
don’t actually have a dielectric. The tr —>■ oo limit corresponds to
a conductor. In that case, both U and F are zero. Basically, all of
the charge on the plates shifts to the overlap x region, and compen¬
sating charge gathers there in the dielectric, so in the end there is no
field anywhere. Note that F decreases as x increases. You should think
about why this is the case. Hint: First convince yourself why the force
746
Solutions to the problems
(a) d
l
O
T 9
o
(b) d
O-•-o-•
i l
Figure 12.129.
should be proportional to the product of the charge densities (and not
the total charges) on the two parts of the plates. And then look at
Exercise 10.15.
10.3 Energy of dipoles
The first configuration is shown in Fig. 12.129(a). There are four relevant
(non-internal) pairs of charges, so the potential energy is (with i d)
q2 )
Vd 2 + i 2 )
*-(x- , 1
4xe Q d ^ v 7 ! +l 2 /d 2
2 q 2
47T60 d
q 2 l 2
4jreo4 3 4 jt6o d 2
(12.457)
where we have used l/\/l + e ^ 1 — e/2. The second configuration is
shown in Fig. 12.129(b). The potential energy is now
4tt6o
4
r
4 ne^d
q 2
d-l d+l
47T60 d
1
1
1 -t/d l+l/d
4neo d \ d 2
2 -
21‘
, l
+ d + d 2
l l 2
d + d 2
2neQd 2
(12.458)
where we have used 1/(1 + e) Ri 1 — e + e 2 . Note that we needed to
go to second order in the Taylor expansions here. By looking at the initial
expressions for U for each setup, it is clear why the first U is positive, but
not so clear why the second U is negative. However, in the limit where the
dipoles nearly touch, the second U is certainly negative.
10.4 Dipole polar components
Remember that our convention for the angle 9 is that it is measured down
from the z axis in Fig. 10.6. So the radial unit vector is given by ? =
sin 9 x + cos 9 z. The tangential unit vector, which is perpendicular to r, is
then given by 6 = cos 9 x — sin 9 z; this makes the dot product of r and 6
equal to zero, and you can check that the overall sign is correct. Inverting
these expressions for f and 8 gives
x = sin 9 r + cost? 0 and z = cos 9 r — sin# 0. (12.459)
Therefore,
E = £ a x + E z z
= F/Tsin# r + cos 9 0)+ E z (cos9 r — sinO 8)
= r{E x sin@ + E z cos 9) + 8{E X cos 9 — E z sin 6?)
= ——— T (f [(3 sin 9 cos 9) sin 9 + (3 cos 2 9—1) cos 9
47reor j V
+ 0r(3sin0cos6 l )cos@ — (3cos 2 $ — l)sin@] ).
(12.460)
12.10 Chapter 10
747
Using sin 2 # + cos 2 # = 1 in the r term, E quickly simplifies to
E = ——-(2cos#r + sin#0), (12.461)
4ne 0 r 3 v ’
as desired. Alternatively, E r equals the projection of E = ( E x , E z ) onto
r = (sin 9, cos 9). Since r is a unit vector, this projection equals the dot
product E • r. Therefore,
E r = E ■ r = (E x , E z ) ■ (sin 9, cos 8) = E x sin 8 + E, cos 8, (12.462)
in agreement with the third line in Eq. (12.460). Likewise,
Eg = E 8 = (E X ,E : ) ■ (cos#, — sin#) = £ A cos# - E, sin#, (12.463)
again in agreement with the third line in Eq. (12.460).
10.5 Average field
(a) From part (c) of Problem 1.28 we know that the average electric field
over the volume of a sphere of radius R, due to a given charge q at
radius r < R, has magnitude qr/4nef)R 3 and points toward the center
(if q is positive). In vector form, this average field can be written as
— qr/4neQR 3 . If we sum this over all the charges inside the sphere,
then the numerator becomes #/ r i ( or / r P dv if we have a contin¬
uous charge distribution). But this sum is, by definition, the dipole
moment p, where p is measured relative to the center. So the average
field over the volume of the sphere is E avg = — p/Ane^R 3 , as desired.
Note that all that matters here is the dipole moment; the monopole
moment (the total charge) doesn’t come into play.
(b) Since E avg is proportional to 1 /R 3 , and since volume is proportional
to R 3 , the total integral of E over the volume of a sphere is indepen¬
dent of R (provided that R is large enough to contain all the charges).
This means that if we increase the radius by dR, we don’t change the
integral of E. This implies that the average value of E over the sur¬
face of any sphere containing all the charges equals zero. (We actu¬
ally already knew this from part (a) of Problem 1.28. Each individual
charge yields zero average field over the surface.) A special case of
this result is the centered point-dipole case in Exercise 10.25.
So for the specific case shown in Fig. 10.32(a), the average value
of the field over the surface of the sphere is zero. And since the
dipole moment has magnitude p = 2 ql and points upward, the result
from part (a) tells us that the average value over the volume of the
sphere, E av g = — p/4jreqE 3 , has magnitude ql/lne^R 3 and points
downward.
(c) The average value of the field over the surface of the sphere in
Fig. 10.32(b) is not zero. From part (b) of Problem 1.28, the aver¬
age field due to each charge has magnitude q/Ane^l 3 and points
downward. So the average field over the surface, due to both charges,
has magnitude q/lne^l 3 and points downward. Since this is inde¬
pendent of the radius of the sphere, the average field over the vol¬
ume of a sphere with R<£ also has magnitude q/lne^l 3 and points
downward.
748
Solutions to the problems
The moral of all this is that “outside” the dipole, the field points in
various directions and averages out over the surface of a sphere. But
“inside” the dipole, the field points generally in one direction, so the
average is nonzero over the surface of a sphere.
Note that volume average of E is continuous as R crosses the R = l
cutoff between the two cases in parts (a) and fb); in both cases it has
magnitude q/2neol 2 . If we multiply this by l/l and use p = ql, we
can write it as p/lne^. Multiplying by the volume Anl 2 /3 then
tells us that the total volume integral of E, over a sphere of radius
l, has magnitude 2p/3eo and points downward. In other words, for
a fixed value of p, even the limit of an idealized dipole still has a
nonzero value of f E dv, despite the fact that the only shells yielding
nonzero contributions are infinitesimal ones.
10.6 Qucidrupole tensor
Our goal is to find the potential <p( r) at the point r = ( xi,x 2 ,x 3 ). As
in Section 10.2, primed coordinates will denote the position of a point
in the charge distribution. The distance from r to a particular point r' =
(XpxJ,,Xj) in the distribution is
R =
= yj Co - td x ) 2 + (x 2 - 4) 2 + (x 3 - y 3 ) 2
- r ' /1 + -2 ~
r- 2
= o 1 +
/2 _ 2^x ; /.
r 2 r
(12.464)
where we have used 53 = i~ and 53-'f 2 = r,2 > an d where (xi,X 2 ,x 3 ) =
(xj,X 2 ,X 3 )/r is the unit vector r in the r direction. Assuming that r' is
much smaller than r, we can use the expansion (1 + 5) -1 / 2 = 1 — 8/2 +
3<5 2 /8 — • • • to write (dropping terms of order 1/r 4 and higher)
1
R
1
r
1
r
3 (E ^.) 2
2r2
3(E^.) 2
2r 2
Z 2 '
2 r 2
(ZxjV 2 '
2 r-
(12.465)
In the last term here, we have multiplied by 1 in the form of the square of
the length of a unit vector, for future purposes. It is easier to understand
this result for 1 /R if we write it in terms of vectors and matrices:
1 1 I „ „ „
S = - + — (xi,x 2 ,x 3 )
R r r z
1 „ „ „
+ ^3(xi,x 2 ,x 3 )
/ 3x ' 2 - / 2
3x^Xj
v 3xjXj
3^2 - r' 2
3x^x(,
(12.466)
Sx^Xj ^
3x^Xj
3x! 2 - Z 1 )
12.10 Chapter 10
749
You should verify that this is equivalent to Eq. (12.465). If desired, the
diagonal terms of this matrix can be written in a slightly different form.
Since r'~ = x /2 + a /, 2 + x^ 2 , the upper left entry equals 2x /2 — x ^, 2 — a/, 2 .
Likewise for the other two diagonal entries. Note that there are only five
independent entries in the matrix, because it is symmetric and has trace
zero.
To obtain <p(r), we must compute the integral.
<t>{ r) =
1 f p(r')dv'
47T£ 0 J R
(12.467)
In other words, we must compute the volume integral of Eq. (12.466) times
p(r'), and then tack on a l/4jreo- When the 1/r term is integrated, it
simply gives q/r, where q is the total charge in the distribution. To write
the other two terms in a cleaner way, define the vector p to be the vector
whose entries are the p dv' integrals of the entries in the above (,r / |, Xj , x^)
vector. And likewise define the matrix Q to be the p dV integral of the
above matrix. For example, the first component of p and the upper-left
entry of Q are
PI =
= j x\p(r')dv
and
fill =
= f (3xj 2 — r' 2 )p{r')dv ,
(12.468)
and so on. We can then write the result for the potential at an arbitrary
point r in the compact form.
<t>( r) =
1
4rr6o
f ■ Qr
(12.469)
The advantage of Eq. (12.469) over Eq. (10.9) in the text is the following.
The latter gives the correct value of cp at points on the z axis. However, if
we want to find <p at another point, we must redefine 9 as the angle with
respect to the direction to the new point, and then recalculate all the K
The present result in Eq. (12.469) has the benefit that, although it involves
calculating a larger number of quantities, it is valid for any choice of the
point r. The quantities q, p, and Q depend only on the distribution, and
not on the point r at which we want to calculate the potential. Conversely,
the quantities r and r in Eq. (12.469) depend only on r and not on the
distribution. So, for a given charge distribution, we can calculate (with
respect to a given set of coordinate axes) p and Q once and for all. We then
simply need to plug our choice of r into Eq. (12.469), and this correctly
gives <p{ r) up to order 1 /r 5 .
In the special case where r lies on the z = x 3 axis, we have r =
(0,0. 1). Since only X 3 is nonzero, only £>33 (the lower right entry in
Q) survives in the dot product f • Qr. Furthermore, if 9 is the angle of
r' with respect to the X 3 axis, then we have x^ = /cos#. So £>33 =
/ r ,2 (3 cos 2 9—l)p dv'. When the l/2r 5 factor in Eq. (12.469) is included,
we correctly arrive at the result Eq. (10.9).
For a spherical shell, which we know has only a monopole moment,
you can quickly verify that all of the entries in Q are zero. The off-diagonal
t H
750
Solutions to the problems
entries are zero from symmetry, and the diagonal elements are zero due
to the example in Section 10.2 combined with the previous paragraph.
Alternatively, the average value of, say, x /2 over the surface of a sphere
equals r ,2 /3, because it has the same average value as x(, 2 and x^ 2 , and the
sum of all three averages is r /2 . If you want to get some practice with Q,
Exercise 10.26 deals with the quadrupole arrangement in Fig. 10.5.
10.7 Force on a dipole
Let the dipole consist of a charge — q at position r and a charge q at posi¬
tion r + s. Then the dipole vector is p = qs. If the dipole is placed in an
electric field E, the net force on it is
F= (- 9 )E(r) + ?E(r + s).
(12.470)
The x component of this isF* = (—q)E x (r) + qE x (r+s). Now, the change
in a function/ due to a small displacement s is V/ • s, by the definition of
the gradient (or at least that’s one way of defining it). So we can write F x as
F x = q[E x (r + s) - E x ( r)] = qVE x ■ s
= (qs) ■ VF* = p • VE X ,
(12.471)
as desired. Likewise for the other two components.
10.8 Force from an induced dipole
If q is the charge of the ion, then the magnitude of the electric field of the
ion at the location of the atom is E = q/4neor 2 . If the polarizability of
the atom is a, then the induced dipole moment of the atom is p — aE =
aq/4neo>~. This dipole moment points along the line from the ion to the
atom (see Fig. 12.130), so the magnitude of the field of the induced dipole
at the location of the ion is i?dipole = 2p/4jreo'' 3 . The magnitude of the
force on the ion is therefore
l — ^/Tdipole — n
4neQ
You can quickly show that the force is attractive for either sign of q. The
potential energy relative to infinity is
(12.473)
The polarizability of sodium is given by a/47i€Q = 27 ■ 10 -30 m 3 . If the
magnitude of the potential energy equals \U\ =4 - 10~ 21 J, then solving
for r and setting q — e gives
1/4
(27 • l(r 30 m 3 )(1.6 • 10“ 19 C) 2
P
r =
= 9.4- 10“ 10 m.
(12.474)
If r is larger than this, then (on average) the thermal energy is sufficient to
Figure 12.130.
kick the ion out to infinity.
12.10 Chapter 10
751
10.9 Polarized water
We must determine the number, n, of molecules of water per cubic cen¬
timeter. A mole of something with molecular mass M has a mass of M
grams. (Equivalently, since the proton mass is 1.67 ■ 10 -24 g, it takes
1/(1.67 • 10~ 24 ) = 6- 10 23 protons to make 1 gram, and this number is
essentially Avogadro’s number.) Water has a molecular weight of 18, so
the number of water molecules per gram (= cm 3 ) is n = (6 • 10 23 /mole)/
(18 cm 3 /mole) = 3.33 ■ 10 22 cm -3 . The dipole moment of water can be
written as p = 6.13 ■ 10 -28 C-cm. Assuming the dipoles all point down,
the polarization density is therefore
P = np = (3.33 ■ 10 22 cm _3 )(6.13 ■ 10 -28 C-cm) = 2.04 • 10 -5 C/cm 2 .
(12.475)
From the reasoning in Section 10.7, this is the surface charge density,
a. The number of electrons per square centimeter it corresponds to is
ale = (2.04- 10“ 5 C/cm 2 )/(1.6 • 10“ 19 C) = 1.3 ■ 10 14 cm“ 2 . This is
somewhat smaller than the number of surface molecules per square cen¬
timeter, which equals n 2 / 3 = 1.0 • 10 15 cm -2 because each edge of the
1 cm 3 cube is (approximately) n 1 / 3 molecules long.
10.10 Tangent field lines
Consider the Gaussian surface indicated by the heavy line in Fig. 12.131.
The side part of the surface is constructed to follow the field lines, so
there is no flux there. Likewise, there is no flux through the top circular
face, because the field is zero outside the capacitor plates. So the only flux
comes from the great circle inside the sphere. From Eq. (10.53) the field
inside the sphere has the uniform value of 3 Eq/(2 + k). So the flux out of E = 0
the Gaussian surface equals —nR~ ■ 3 Eq/(2 + k), where the minus arises _
because the flux is inward.
The total charge enclosed in the Gaussian surface comes from two
places: the negative charge in the circle on the upper capacitor plate,
and the positive charge on the upper hemisphere. The former is simply
?cap = (—a)nn = (— e^E^Tcr 2 , where we have used the fact that the
charge densities on the capacitor plates are what cause the uniform field
Eq', hence Eq = cr/e q- The latter charge is just q sp h = PnR~, where P is
the polarization, because the top patch of the column in Fig. 10.21(a) has
a charge of Pda (where da is the horizontal cross-sectional area), inde¬
pendent of the tilt angle of the actual end face. And all the da areas sim¬
ply add up to the great-circle area, nR 2 . (Or you could just integrate the
P cos 6 surface density over the hemisphere.) Using the value of P from
Eq. (10.54), Gauss’s law gives
^ — — (?cap + 9sph)
k + 2
r = R
(12.476)
Figure 12.131.
752
Solutions to the problems
(Inside S)
(Outside S)
O-
o-
. o
o-
o-
o-
! O
o
o-
-9
Figure 12.132.
As a check, we have r = R when k = 1. In this case, our dielectric is
just vacuum, so the field remains Eg everywhere; the field lines are all
straight. Also, we have r = V3 R when k —>■ oo. In this limit the sphere
is a conductor. The factor of a/ 3 isn’t so obvious. Note that, in the case of
a conductor, a field line can’t actually be tangent to the surface, because
field lines must always be perpendicular to the surface of a conductor.
What happens is that the external field approaches zero along the equator
(the zero vector is, in some sense, both parallel and perpendicular to the
surface). But a tiny distance away from the equator, the field is nonzero, so
it is meaningful to ask where that field line ends up on the distant capacitor
plates.
10.11 Bound charge and divergence of P
If we take the volume integral of both sides of Eq. (10.61) and use the
divergence theorem, we see that our goal is to show that f s P ■ da —
— Abound > where Abound * s the bound charge enclosed within the surface S.
Assume that the polarization P arises from N dipoles per unit vol¬
ume, each with a dipole moment p = qs. Then P = N p = Nqs. If the
dipoles point in random directions, so that P = 0, then there is no extra
bound charge in a given volume. But if they are aligned, so that P ^ 0,
and if additionally P varies with position, then there may be a net bound
charge in the volume. The reasoning is as follows.
Consider a collection of dipoles, as shown in Fig. 12.132. The vertical
line represents a patch of the right-hand surface of S. How much extra
negative charge is there inside S, that is, to the left of the line? If a given
dipole lies entirely inside or outside S, then it contributes nothing to the
net charge. But if a dipole is cut by the vertical line, then there is an extra
charge of — q inside S.
How many dipoles are cut by the line? Any dipole whose center lies
within s/2 of the line gets cut by it. So the center must lie in a slab with
thickness s, indicated by the shaded region in the figure. The two extreme
dipole positions are indicated by the boxes. If the area of a given patch of
the surface is da, then any dipole whose center lies in a slab of volume
sda will contribute a charge of —q to 5. Since there are N dipoles per unit
volume, we see that N(sda) dipoles are cut by the line. The extra charge
associated with the patch is therefore Abound = N(s da)(—q), which can
be written as Abound = ~(Nqs)da = —Pda.
If a dipole is tilted at an angle 6 with respect to the normal to the
patch, then the volume of the relevant slab is decreased by a factor of
cos 9. If we tack this factor onto P, it simply turns P into the component
P l perpendicular to the surface. So in general the extra charge inside the
volume, near a given patch with area da, equals Abound = ~P±da, which
can be written as the dot product, Abound = — P ■ da. Integrating this over
the entire surface gives the total enclosed bound charge as
Abound = - J P da, (12.477)
as desired.
Although we motivated this result in Section 10.11 by considering
dielectrics, this problem shows (as mentioned in the text) that this result is
12.10 Chapter 10
753
quite independent of dielectrics. No matter how the polarization P comes
about, the result in Eq. (12.477) is still valid. (You can manually twist the
dipoles in whatever way you want, provided that P changes slowly on the
length scale of the dipoles, so that we can talk about smooth averages.)
To emphasize what we said in the text, the logical route to Eq. (10.62) is
to start with Eqs. (10.59) and (10.61), both of which are universally true,
and then Eq. (10.62) immediately follows. No mention has been made of
dielectrics. But if we are in fact dealing with a (linear) dielectric, then
P = X(YoE’ and we can use 1 + Xe = k 1° say that additionally
6qE + P = cqE + Xe<T)E = kcoE = eE.
(12.478)
In all cases the relation D = gpE + P holds, but that is just a definition.
10.12 Boundary conditions for D
£>j_ is continuous. This follows from div D = pf ree ; there is no free charge
in the setup, so the divergence of D is zero. The divergence theorem then
tells us that f D • cla = 0 for any closed surface. That is, there is zero
flux through any surface. So if we draw a pancake-like pillbox with one
face just inside the slab and one face just outside, the inward flux through
one face must equal the outward flux through the other. Hence D'^A =
D'j^A => = ZYj 01 . That is, D± is continuous across the boundary.
For £>|j , we know that £y is continuous across the boundary, because
all we have at the boundary is a layer of bound charge, which produces no
discontinuity in Fy . So D = epE + P tells us that the discontinuity in D y is
the same as the discontinuity in Py . Since P = 0 outside, the discontinuity
in Py is simply — Pj". That is, the change in Dy when going from inside to
outside is — Pj".
10.13 Q for a leaky capacitor
From Exercise 10.42, the energy density in the electric field is eE 2 / 2.
And it is the same for the magnetic field, by plugging B = ^/JZ^eE into
B 2 / 2/zp. The total energy density is therefore eE 2 , or eE^ cos 2 cot. But the
time average of cos 2 cot is 1 /2, so the average energy density is eE^/2.
The energy in the fields will decay due to ohmic resistance. To cal¬
culate this power dissipation, consider a tube of cross-sectional area A and
length L. The power dissipated in this tube is
P = I 2 R = ( JA) 2 (pL/A) = J 2 p(AL)
2 1 9
= (&E) — (volume) = a# (volume).
(12.479)
o
The power dissipated per unit volume is therefore oE 2 . The time average
of this is oEq/2. Hence
Q =
co ■ (energy stored) a>(eE^/2) coe
power loss
a
(12.480)
754
Solutions to the problems
as desired. From Table 4.1, the conductivity of seawater is a =
4 (ohm-m) -1 . And from Fig. 10.29, the dielectric constant k is still about
80 at a frequency of 1000 MHz (10 9 Hz). Therefore, since e = kcq,
we have
Q =
(2:x • 10 9 s -1 )(80 • 8.85 • 10“ 12
4 (ohm-m) -1
(12.481)
Since Q equals the number of radians of cot required for the energy to
decrease by a factor of 1/e, we see that by the end of one cycle (2 n radi¬
ans) there is practically no energy left. The wavelength corresponding to
1000 MHz is (c/ *Jk)/v = 0.033 m. So microwave radar won’t find sub¬
marines !
10.14 Boundary conditions on E and B
With no free charges or currents, the equations describing the system are
V • D = 0, V x E = —3B/3t;
V ■ B = 0, VxB = /z 0 3D/3 1 . (12.482)
The two equations involving D come from Eqs. (10.64) and (10.78) with
Pfree and Jfree set equal to zero. The other two equations are two of
Maxwell's equations. We can now apply the standard arguments. For the
perpendicular components, we can apply the divergence theorem to the
two “div” equations, with the volume chosen to be a squat pillbox, of van¬
ishing thickness, spanning the surface. Our equations tell us that the net
flux out of the volume is zero, so the perpendicular field on one side must
equal the perpendicular field on the other. And for the parallel compo¬
nents, we can apply Stokes’ theorem to the two “curl” equations, with the
area chosen to be a thin rectangle, of vanishing area, spanning the surface.
Our equations tell us that the line integral around the rectangle is zero,
so the parallel field on one side must equal the parallel field on the other.
(The finite non-zero entries on the right-hand sides of the curl equations
are inconsequential, because they provide zero contribution when inte¬
grated over the area of an infinitesimally thin rectangle.) The above four
equations therefore yield (with 1 and 2 labeling the two regions)
= D2,_L> e \,\\= e 1 ,\\\
5l,±=-B 2 ,-L, B l.l =B U- (12.483)
Since D = eE for a linear dielectric, the first of these equations gives
ei E l,X = e 2 E 2 ,_L- (12.484)
So E± is discontinuous. But the other three components are continuous
across the boundary. That is, the entire B field is continuous, as is the
parallel component of E.
Note that we are assuming that the materials aren’t magnetic. After
reading Section 11.10, you can show that in magnetic materials there is a
discontinuity in B\\.
12.11 Chapter 11
755
12.11 Chapter 11
11.1 Maxwell’s equations with magnetic charge
Maxwell’s equations with only electric charge and electric current are
given in Eq. (9.17). If magnetic charge existed, the last equation would
have to be replaced, as discussed in Section 11.2, by V • B = b\ri, where
fj is the magnetic charge density, and b\ is a constant that depends on
how the unit of magnetic charge is chosen. With the conventional defini¬
tion of the direction of B, a positive magnetic charge would be attracted
to the north pole of the earth, so it would behave like the north pole of a
compass.
Magnetic charge in motion with velocity v would constitute a mag¬
netic current. Let K be the magnetic current density. Then K = ?;v, in
analogy with J = p\. Conservation of magnetic charge would then be
expressed by the “continuity equation,” V • K = — dq/dt, in analogy with
V ■ J = -3 p/dt.
A magnetic current would be the source of an electric field, just as
an electric current is the source of a magnetic field. So we must add to the
right side of the first Maxwell equation in Eq. (9.17) a term proportional
to K. (Equivalently, if we didn’t add such a term, we would end up with a
contradiction, similar to the one in Section 9.1, arising from the fact that
V • (V x E) = 0 is identically zero.) Let this new term be & 2 K. Then
we have
9B
V x E =-+ 3iK. (12.485)
3 1
To determine the constant b 2 , we can take the divergence of both sides of
this equation. The left-hand side is identically zero because V • (V x E) =
0, so we have (using the continuity equation)
/ 3B\
0 =- v -U) +fo2V - K
3 /v7 t f dq
3 7
(12.486)
Therefore £>2 must equal —b\. So the generalized Maxwell’s equations
take the form (with b = b\ = —b 2 ),
3 /
3 , drj
=- (b]T]) — 3i —
dt 1 - dt
dri
= -{b\ + b 2 ) — .
3B
V x E =- bK,
dt
3E
V x B = p 0 e 0 — + mo J,
p
V ■ E = —,
e 0
V • B = bi].
(12.487)
The constant b can be chosen arbitrarily. Two common conventions are
3=1 and b = mo-
756
Solutions to the problems
m sin 9
11.2 Magnetic dipole
If we treat the current loop like an exact dipole, then the dipole moment
is m = la = Inb 2 . Equation (11.15) gives the magnetic field at position z
along the axis of the dipole as pQm/2nz 2 , which here equals pq{I nb 2 )/
2nz 3 = p-olb 2 / 2 z 3 .
If we treat the current loop (correctly) as a loop of finite size, then
Eq. (6.53) gives the field at position z on the axis as B z = /xq/ft 2 /2(z 2 +
ft 2 ) 3 /-. For z^> b we can ignore the ft 2 term in the denominator, yielding
B z Ri pq /ft 2 /2z 3 , which agrees with the above result for the idealized
dipole.
The correct result is smaller than the idealized-dipole result by the
factor z 3 /(z 2 + ft 2 ) 3 / 2 . This factor approaches 1 as z —> oo. It is larger
than a given number rj (we are concerned with r] = 0.99) if
Z 3 Z 2 2/3 ’l X ^b
(z 2 + ft 2 ) 3 / 2 Z 2 + ft 2 yi - p 2 / 3
(12.488)
For g = 0.99 this gives z > (12.2)ft. You can show that if we want the
factor to be larger than 1 — e (so e = 0.01 here), then to a good approx¬
imation (in the limit of small e) we need z/ft > y/3/2e. And indeed,
^3/2(0.01) = \/l50 = 12.2.
11.3 Dipole in spherical coordinates
Using the V x (A x B) vector identity from Appendix K, with m constant,
we find (ignoring the po/4: r for now)
B oc V x [m x (r/r 2 )] = m(V • (r/r 2 )) - (m • V)(r/r 2 ). (12.489)
But the divergence of r/r 2 is zero (except at r = 0), because we know
that the divergence of the Coulomb field is zero; alternatively we can just
use the expression for the divergence in spherical coordinates. So we are
left with only the second term. Therefore, using the expression for V in
spherical coordinates,
/ 3 13) r
B oc — ( +W 0 -— ) -j. (12.490)
\ or r 30 J r z
In the 3/3 r term here, the vector r doesn’t depend on r, but r 2 does, of
course, so m r (d/dr)(r/r 2 ) = — 2m,r/r 3 . In the 3/3 9 term, r 2 doesn’t
depend of 9, but the vector f does. If we increase 9 by d9, then r
changes direction by the angle d6. Since f has length 1, it therefore
picks up a component with length d9 in the 0 direction. See Fig. F.3 in
Appendix F; that figure is relevant to the oppositely defined 9 in cylindrical
coordinates, but the result is the same. Hence 3r/3 9 = 8. So we have
( mg/r)(d/d9)(r/r 2 ) = mgO/r 3 .
Finally, in Fig. 12.133 we see that the components of the fixed vec¬
tor m = mx relative to the local r-6 basis are m r = m cos 9, and mg =
—m sin ft. The negative sign here comes from the fact that m points
12.11 Chapter 11
757
partially in the direction of decreasing 9 (at least for the right half of the
sphere). Putting this all together, and bringing the ixq/Ajt back in, gives
B = -
MO
47T
. Mo'»
2 n r 3
r 6
—2(mcos9)^r + (-msinfl)^
r 5 r 5
a , a^ m ■ a
cos 9 + 9 -- sintf,
4jrr 3
(12.491)
11.4
in agreement with Eq. (11.15).
Force on a dipole
(a) The expression (m • V)B is shorthand for
(m ■ V)B =
'^ + m y -+m z -)(B x ,By,B z x
(12.492)
The operator in parentheses is to be applied to each of the three com¬
ponents of B, generating the three components of a vector. In the setup
in Fig. 11.9 with the ring and diverging B field, m z is the only nonzero
component of m. Also, B x and B v are identically zero on the z axis,
so dBx/dz and 3 B v /dz are both zero (or negligibly small close to the
z axis). Therefore only one of the nine possible terms in Eq. (12.492)
survives, and we have
(m • V)B =
0 , 0 , m 7
3 B z
3 z
(12.493)
as desired.
(b) The expression V (m • B) is shorthand for
( 3 3 3 \
—, —, — J (m x B x + m y By + m z B z ). (12.494)
Each derivative acts on the whole sum in the parentheses. But again,
only m- is nonzero. Also, on the z axis, B z doesn’t depend on x or
y, to first order (because, by symmetry, B z achieves a maximum or
minimum on the z axis, so the slope as a function of x and y must be
zero). Hence dB z /dx and dB z /dy are both zero (or negligibly small
close to the z axis). So again only one term survives and we have
V(m • B) = U),0, . (12.495)
as desired.
(c) Let’s first see what the two expressions yield for the force on the given
square loop. Then we will calculate what the force actually is. The
dipole moment m points out of the page with magnitude /(area), so
we have m = zla 2 . Using the above expressions for (m • V)B and
V(m • B) in Eqs. (12.492) and (12.494), we obtain
0 + 0 + (fa 2 )^-j (0,0, Bo*) = (0,0,0) (12.496)
(m • V)B =
758
Solutions to the problems
and
V(m • B) = A I'j (0 + 0 + (/a 2 )B 0 x) = (/a 2 B 0 ,0,0).
(12.497)
We see that the first expression yields zero force on the loop, while
the second yields a force of /a 2 Bo in the positive x direction.
Let’s now explicitly calculate the force. We quickly find that the
net force on the top side of the square is zero (the right half cancels
the left half). Likewise for the bottom side. Alternatively, the corres¬
ponding pieces of the top and bottom sides have canceling forces.
So we need only look at the left and right sides. By the right-hand
rule, the force on the right side is directed to the right with magnitude
IBl = /(Bofl/2)(a) = IBqct /2. The force on the left side also points
to the right (both I and B switch sign) with the same magnitude. The
total force is therefore F = /Bqu 2 in the positive x direction, in agree¬
ment with Eq. (12.497). So V(m • B) is the correct expression for the
force. (Actually, all that we’ve done is rule out the (m • V)B force. But
V (m • B) is in fact correct in all cases.)
11.5 Converting Xm
Consider a setup in which the SI quantities are M = 1 amp/m and B = 1
tesla. Then Xm — Mo44/B = 47r ■ 10 -7 . You can verify that the units do
indeed cancel so that Xm is dimensionless.
How would someone working with Gaussian units describe this
setup? Since 1 amp/m equals (3 ■ 10 9 esu/s)/(100cm), this would be the
value of M in Gaussian units if there weren’t the extra factor of c in the
definition of m. This factor reduces the value of all dipole moments m (and
hence all magnetizations M) by 3 ■ 10 10 cm/s. The value of M in Gaussian
units is therefore
M =
3 • 10 9 esu/s 1
100 cm 3 • 10 10 cm/s
10 “ 3
esu
cm 2
(12.498)
Both of the factors of 3 here are actually 2.998, so this result is exact.
The magnetic field in Gaussian units that corresponds to 1 tesla is
10 4 gauss, so the susceptibility in Gaussian units for the given setup is
M 10 3 esu/cm 2
B 10 4 gauss
10
-7
= 10
-7
cnL gauss
(12.499)
The units do indeed cancel, because the expression for the Lorentz force
tells us that a gauss has the units of force per charge. So the units of Xm are
esu 2 / (cm 2 • force). And these units cancel, as you can see by looking at the
units in Coulomb’s law. The above value of Xm in SI units was 47r ■ 10“ 7 ,
which is 4 n times the Gaussian value, as desired.
11.6 Paramagnetic susceptibility of liquid oxygen
Equation (11.20) gives the force on a magnetic moment as F = m(dB z /dz).
Using the data in Table 11.1, and taking upward to be positive for all
quantities, the magnetic moment of a 10 -3 kg sample is
12.11 Chapter 11
759
F -7.5 • 10“ 2 N
= 4.4- 10 -3 J/T. (12.500)
m =
dB-Jdz —17 T/m
The magnetic susceptibility is defined via M = XmB/Po- (The accepted
M = XniH definition would give essentially the same result, because Xm
will turn out to be very small. See Exercise 11.38.) The volume of 1 gram
of liquid oxygen is V = (10 -3 kg)/(850 kg/m 3 ) = 1.18 • 10 -6 m 3 . So
M ( m/V ) m/XQ
B/p 0 B/n o BV
(4.4 • 10“ 3 J/T) (4;r ■ 10 -7 kg m/C 2 )
(1.8 T) (1.18 ■ 10 -6 m 3 )
= 2.6 ■ 10 -3 . (12.501)
11.7 Rotating shell
For the magnetized sphere, we know from Eq. (11.55) that near the equa¬
tor the surface current density is equal to M, because the sphere looks
essentially like a cylinder there (the surface is parallel to M). But away
from the equator, the surface is tilted with respect to M. From the exam¬
ple at the end of Section 11.8, the surface current density is given by
J = M || => J (9) = Ms'm9, where 9 is the angle down from the
top of the sphere (assuming that M points up).
Now consider a rotating sphere with uniform surface charge den¬
sity a . The surface current density at any point is J = ov, where v =
cd (R sin$) is the speed due to the rotation. Hence J(9) = ocoR sin 9. The
J(9) expressions for the magnetized and rotating spheres have the same
functional dependence on 6, so they will be equal for all 9 provided that
M = aajR.
11.8 B inside a magnetized sphere
(a) The field in Eq. (11.15) is obtained from the field in Eq. (10.18) by
letting p — > m and eg “► 1/Mo- If we replace all the electric dipoles
p in a polarized sphere with magnetic dipoles m, then at an external
point, the field from each dipole is simply multiplied by (m/p)(p oH)).
The integral over all the dipole fields is multiplied by this same factor,
so the new magnetic field at any external point equals (m/p)(p. oeo)
times the old electric field. We know from Section 10.9 that the old
external electric field is the same as the field due to an electric dipole
with strength pq = (4nR^/3)P, with P = Np, located at the center.
You can quickly check that (m/p)(p, o 6 o) times this field is the same
as the magnetic field due to a magnetic dipole with strength in q =
(4jtR 3 /3)M, with M = Nm.
(b) If mg points in the z direction, then from Eq. (11.12) the Cartesian
components of A at points (x, y, z) on the surface of the sphere are
M o mgy My
Pq mgx Mx
(12.502)
760
Solutions to the problems
Note that the result from Problem 11.7 then tells us that the A on
the surface of a spinning spherical shell equals (p,QOcoR/3)(—y,x, 0).
This agrees with the A we found in a different manner in Problem 6.7.
Recall from Section 6.3 that A x satisfies V 2 A_ v = —/IqJ x . And
similarly for Ay. But J = 0 inside the sphere, so both A r and Ay satisfy
Laplace’s equation there. By the uniqueness theorem, this means that
if we can find a solution to Laplace’s equation inside the sphere that
satisfies the boundary conditions on the surface of the sphere, then we
know that we have found the solution. And just as with the cj> for the
polarized sphere in Section 10.9, the solutions for A x and A y are easy
to come by. They are simply the functions given in Eq. (12.502); their
second derivatives are zero, so they each satisfy Laplace’s equation.
The magnetic field inside the sphere is then
B = V x A =
B-qM
3
* y
d/dx d/dy
-y X
z
a/3 z
o
2(ioM „
-z.
(12.503)
Like the E inside the polarized sphere, this B is uniform and points
vertically. But that is where the similarities end. This B field points
upward, whereas the old E field pointed downward. Additionally, the
numerical factor here is 2/3, whereas it was (negative) 1/3 in E.
The 2/3 is exactly what is needed to make the component normal
to the surface be continuous, and to make the tangential component
have the proper discontinuity (see Exercise 11.31).
Equation (12.503), combined with the result from Problem 11.7,
tells us that the field throughout the interior of a spinning spherical
shell is uniform and has magnitude 2p.QOcoR/3. This is consistent
with the result from Problem 6.11 for the field at the center of the
sphere.
11.9 B at the north pole of a spinning sphere
From Problem 11.7, we know that the magnetic field due to a spinning
shell with radius r and uniform surface charge density a is the same (both
inside and outside) as the field due to a sphere with uniform magnetization
M r = a a>r. And then from Problem 11.8 we know that the external field
of a magnetized sphere is that of a dipole with strength m = (4nr 3 /3)M r
located at the center. So the (radial) field at radius R outside a spinning
shell with radius r (at a point located above the north pole) is
H 0 m (iq 4n r 3 {a air) 2p. 0 oair 4
JtS = - yr = - W - = -o-. ( 1Z. DU4)
2nR 3 2nR 3 3 3 R 3
We can consider the solid spinning sphere to be the superposition
of many spinning shells with radii ranging from r = 0 to r = R, with
uniform surface charge density a = p dr. The north pole of the solid
sphere is outside all of the shells, so we can use the above dipole form of
B for every shell. The total field at the north pole (that is, at radius R) is
therefore
B = /
Jo
K 2(Iq(p dr)(Dr 4 2p.QptL>R 2
3 R 3 ~ 15
(12.505)
15
12.11 Chapter 11
761
This field is 2/5 as large as the field at the center of the sphere; see
Exercise 11.32. In terms of the total charge Q = (47r/? 3 /3)p, we can write
B as B = hq(dQ/\0tiR.
11.10 Surface current on a cube
Equation (11.55) gives the surface current density as J = M. Since the
units of magnetization (J/Tm 3 ) can also be written as A/m, we have J =
4.8 • 10 5 A/m. This current density spans a ribbon that is £ = 0.05 m wide,
so the current is / = Jl = (4.8 • 10 5 A/m)(0.05 m) = 24,000 A.
The dipole moment of the cube is
m — MV = (4.8- 10 5 JT _1 m _3 )(0.05m) 3 = 60 J/T. (12.506)
The field at a distance of 2 meters, along the axis, is given by Eq. (11.15) as
B =
H 0 m _ (An- 10“ 7 kg m/C 2 ) (60 J/T)
2nr 3
, = 1.5 • 10 -6 T, (12.507)
2jr(2m) 3
or 0.015 gauss. This is about 30 times smaller than the earth’s field of
^ 0.5 gauss, so it wouldn’t disturb a compass much.
11.11 An iron torus
From Fig. 11.32, a B field of 1.2 tesla requires an H field of about 120 A/m.
Consider the line integral fH dl around the “middle” circle of the sole¬
noid, with diameter 11 cm. If / is the current in the wire, then NI = 20/ is
the free current enclosed by our circular loop. Therefore,
J H dl — / fre e =!> (120 A/m) ■ n (0.11 m) = 20/ => / = 2.1 A.
(12.508)
Differences
between SI and
Gaussian units
In this appendix we discuss the differences between the SI and Gaussian
systems of units. First, we will look at the units in each system, and then
we will talk about the clear and not so clear ways in which they differ.
A.1 SI units
Consider the SI system, which is the one we use in this book. The four
main SI units that we deal with are the meter (m), kilogram (kg), second
(s), and coulomb (C). The coulomb actually isn’t a fundamental SI unit;
it is defined in terms of the ampere (A), which is a measure of current
(charge per time). The coulomb is a derived unit, defined to be 1 ampere-
second.
The reason why the ampere, and not the coulomb, is the fundamental
unit involving charge is one of historical practicality. It is relatively easy
to measure current via a galvanometer (see Section 7.1). More crudely,
a current can be determined by measuring the magnetic force that two
pieces of a current-carrying wire in a circuit exert on each other (see
Fig. 6.4). Once we determine the current that flows onto an object during
a given time, we can then determine the charge on the object. On the
other hand, although it is possible to measure charge directly via the
force that two equally charged objects exert on each other (imagine two
balls hanging from strings, repelling each other, as in Exercise 1.36), the
setup is a bit cumbersome. Furthermore, it tells us only what the product
of the charges is, in the event that they aren’t equal. The point is that it is
A.1 SI units
763
easy to measure current by hooking up an ammeter (the main component
of which is a galvanometer) to a circuit. 1
The exact definition of an ampere is: if two parallel wires carrying
equal currents are separated by 1 meter, and if the force per meter on one
wire, due to the entirety of the other wire, is 2 ■ 10“ 7 newtons, then the
current in each wire is 1 ampere. The power of 10 here is an arbitrary
historical artifact, as is the factor of 2. This force is quite small, but by
decreasing the separation the effect can be measured accurately enough
with the setup shown in Fig. 6.4.
Having defined the ampere in this manner, and then having defined
the coulomb as 1 ampere-second (which happens to correspond to the
negative of the charge of about 6.24 • 10 18 electrons), a reasonable thing
to do, at least in theory, is to find the force between two 1 coulomb
charges located, say, 1 meter apart. Since the value of 1 coulomb has
been fixed by the definition of the ampere, this force takes on a particular
value. We are not free to adjust it by tweaking any definitions. It happens
to be about 9 • 10 9 newtons - a seemingly arbitrary number, but in fact
related to the speed of light. (It has the numerical value of c 2 * /10 7 ; we
see why in Section 6.1.) This (rather large) number therefore appears
out in front of Coulomb’s law. We could label this constant with one
letter, such as “k,” but for various reasons it is labeled as 1 /4jreo,
with eo = 8.85 • 10 -12 C 2 s 2 kg -1 m -3 . These units are what are needed
to make the right-hand side of Coulomb’s law, F= (l/4jreo)<7i<72A 2 ,
have units of newtons (namely kg ms -2 ). In terms of the fundamental
ampere unit, the units of eo are A 2 s 4 * kg -1 m -3 .
The upshot of all this is that because we made the choice to define
current via the Lorentz force (specifically, the magnetic part of the Lorentz
force) between two wires carrying current /, the Coulomb force between
two objects of charge q ends up being a number that we just have to accept.
We can make the pre-factor be a nice simple number in either one of these
force laws, but not both. The SI system gives preference to the Lorentz
force, due to the historical matters of practicality mentioned above.
It turns out that there are actually seven fundamental units in the
SI system. They are listed in Table A.l. The candela isn’t relevant to
our study of electromagnetism, and the mole and kelvin come up only
occasionally. So for our purposes the SI system consists of essentially
just the first four units.
1 If we know the capacitance of an object, then we can easily measure the charge on it
by measuring the voltage with a voltmeter. But the main component of a voltmeter is
again a galvanometer, so the process still reduces to measuring a current.
2 The Biot-Savart law, which allows us to calculate the magnetic field that appears in
the Lorentz force, contains what appears to be a messy pre-factor, namely
But since /xq is defined to be exactly An • 10 -7 kg m/C 2 , this pre-factor takes on the
simple value of 10 -7 kgm/C 2 .
764
Differences between SI and Gaussian units
Table A.I.
SI base units
Quantity
Name
Symbol
Length
meter
m
Mass
kilogram
kg
Time
second
s
Electric current
ampere
A
Thermodynamic temperature
kelvin
K
Amount of substance
mole
mol
Luminous intensity
candela
cd
A.2 Gaussian units
What do the units look like in the Gaussian system? As with the SI sys¬
tem, the last three of the above units (or their analogs) rarely come up,
so we will ignore them. The first two units are the centimeter and gram.
These differ from the SI units simply by a few powers of 10, so it is easy
to convert from one system to the other. The third unit, the second, is the
same in both systems.
The fourth unit, that of charge, is where the two systems fundamen¬
tally diverge. The Gaussian unit of charge is the esu (short for “electro¬
static unit”), which isn’t related to the coulomb by a simple power of 10.
The reason for this non-simple relation is that the coulomb and esu are
defined in different ways. The coulomb is a derivative unit of the ampere
(which is defined via the Lorentz force) as we saw above, whereas the
esu is defined via the Coulomb force. In particular, it is defined so that
Coulomb’s law,
F = k
qiqir
r 2
(A.I)
takes on a very simple form with k = 1. The price to pay for this simple
form of the Coulomb force is the not as simple form of the Lorentz force
between two current-carrying wires (although it isn’t so bad; like the
Coulomb force in SI units, it just involves a factor of c 2 ; see Eq. (6.16)).
This is the opposite of the situation with the SI system, where the Lorentz
force is the “nice” one. Again, in each system we are free to define things
so that one, but not both, of the Lorentz force and Coulomb force takes
on a nice form.
A.3 Main differences between the systems
In Section A.2 we glossed over what turns out to be the most important
difference between the SI and Gaussian systems. In the SI system, the
constant in Coulomb’s law,
^si
-= 8.988 • 10 9
4jre 0
Nm 2
(A.2)
A.3 Main differences between the systems
765
has nontrivial dimensions, whereas in the Gaussian system the constant
(A.3)
is dimensionless. We aren’t just being sloppy and forgetting to write the
units; k is simply the number 1. Although the first thing that may strike
you about the two k constants is the large difference in their numeri¬
cal values, this difference is fairly inconsequential. It simply changes the
numerical size of various quantities. The truly fundamental and critical
difference is that k si has units whereas k G does not. We could, of course,
imagine a system of units where k = 1 dyne-cm 2 /esu 2 . This definition
would parallel the units of k sl , with the only difference being the numer¬
ical value. But this is not what the Gaussian system does.
The reason why the dimensionlessness of k G creates such a profound
difference between the two systems is that it allows us to solve for the esu
in terms of other Gaussian units. In particular, from looking at the units
in Coulomb’s law, we can write (using 1 dyne = 1 g • cm/s 2 )
(A.4)
The esu is therefore not a fundamental unit. It can be expressed in terms
of the gram, centimeter, and second. In contrast, the SI unit of charge, the
coulomb, cannot be similarly expressed. Since k sl has units of Nm 2 /C 2 ,
the C’s (and everything else) cancel in Coulomb’s law, and we can’t solve
for C in terms of other units.
For our purposes, therefore, the SI system has four fundamental
units (m, kg, s. A), whereas the Gaussian system has only three (cm,
g, s). We will talk more about this below, but first let us summarize the
three main differences between the SI and Gaussian systems. We state
them in order of increasing importance.
(1) The SI system uses kilograms and meters, whereas the Gaussian sys¬
tem uses grams and centimeters. This is the most trivial of the three
differences, because all it does is introduce some easily dealt with
powers of 10.
(2) The SI unit of charge (the coulomb) is defined via the ampere, which
in turn is defined in terms of the force between current-carrying
wires. The Gaussian unit of charge (the esu) is defined directly in
terms of Coulomb’s law. This latter definition is the reason why
Coulomb’s law takes on a nicer form in Gaussian units. The differ¬
ences between the two systems now involve more than simple powers
of 10. However, although these differences can sometimes be a has¬
sle, they aren’t terribly significant. They are just numbers - no differ¬
ent from powers of 10, except a little messier. All of the conversions
you might need to use are listed in Appendix C.
766
Differences between SI and Gaussian units
(3) In Gaussian units, the k in Coulomb’s law is chosen to be dimension¬
less, whereas in SI units the k (which involves eo) has units/' The
result is that the esu can be expressed in terms of other Gaussian
units, whereas the analogous statement is not true for the coulomb.
This is the most important difference between the two systems.
A.4 Three units versus four
Let us now discuss in more detail the issue of the number of units in
each system. The Gaussian system has one fewer because the esu can
be expressed in terms of other units via Eq. (A.4). This has implica¬
tions with regard to checking units at the end of a calculation. In short,
less information is gained when checking units in the Gaussian system,
because the charge information is lost when the esu is written in terms
of the other units. Consider the following example.
In SI units the electric field due to a sheet of charge is given
in Eq. (1.40) as er/2eo- In Gaussian units the field is 2tzo. Recalling
the units of eo in Eq- (1.3), the units of the SI field are kgmC -1 s -2
(or kgmA -1 s -3 if you want to write it in terms of amperes, but we
use coulombs here to show analogies with the esu). This correctly has
dimensions of (force)/(charge). The units of the Gaussian 2no field are
simply esu/cm 2 , but since the esu is given by Eq. (A.4), the units are
g 1//2 cm _1,/2 s _1 . These are the true Gaussian units of the electric field
when written in terms of fundamental units.
Now let’s say that two students working in the Gaussian system are
given a problem where the task is to find the electric field due to a thin
sheet with charge density er, mass m, volume V, moving with a nonrela-
tivistic speed v. The first student realizes that most of this information is
irrelevant and solves the problem correctly, obtaining the answer of 2:to
(ignoring relativistic corrections). The second student royally messes
things up and obtains an answer of cr 3 Vm~ 1 v~ 2 . Since the fundamen¬
tal Gaussian units of a are g 1 / 2 cm -1 / 2 s -i ; the units of this answer are
cr 3 V (g 1 / 2 cm -1 / 2 s -1 ) 3 (cm) 3 g 1 / 2
mv 2 (g)(cm/s) 2 cm 1 / 2 s’
which are the correct Gaussian units of electric field that we found above.
More generally, in view of Eq. (A.4) we see that any answer with the
units of (g 1 / 2 cm -1 / 2 s -1 )(esug -1 / 2 cm -3 / 2 s)” has the correct units for
the field. The present example has n — 3.
There are, of course, also many ways to obtain incorrect answers in
the SI system that just happen by luck to have the correct units. Correct¬
ness of the units doesn’t guarantee correctness of the answer. But the
3
To draw a more accurate analogy: in SI units the defining equation for the ampere
(from which the coulomb is derived) contains the dimensionful constant /iq in the
force between two wires.
A.5 The definition of B
767
point is that because the charge information is swept under the rug in
Gaussian units, we have at our disposal the information of only three
fundamental units instead of four. Compared with the SI system, there is
therefore a larger class of incorrect answers in the Gaussian system that
have the correct units.
A.5 The definition of B
Another difference between the SI and Gaussian systems of units is the
way in which the magnetic field is defined. In SI units the Lorentz force
(or rather the magnetic part of it) is F = gv x B, whereas in Gaussian
units it is F = ( q/c)\ x B. This means that wherever a B appears in
an SI expression, a B/c appears in the corresponding Gaussian expres¬
sion (among other possible modifications). Or equivalently, a Gaussian B
turns into an SI cB. This difference, however, is a trivial definitional one
and has nothing to do with the far more important difference discussed
above, where the esu can be expressed in terms of other Gaussian units.
In the Gaussian system, E and B have the same dimensions. In the
SI system they do not; the dimensions of E are velocity times the dimen¬
sions of B. In this sense the Gaussian definition of B is more natural,
because it makes sense for two quantities to have the same dimensions if
they are related by a Lorentz transformation, as the E and B fields are;
see Eq. (6.76) for the SI case and Eq. (6.77) for the Gaussian case. After
all, the Lorentz transformation tells us that the E and B fields are sim¬
ply different ways of looking at the same field, depending on the frame
of reference. However, having a “cB” instead of a “B” in the SI Lorentz
transformation can’t be so bad, because x and t are also related by a
Lorentz transformation, and they don’t have the same dimensions (the
direct correspondence is between x and ct). Likewise for p and E (where
the direct correspondence is between pc and E). At any rate, this issue
stems from the arbitrary choice of whether a factor of c is included in the
expression for the Lorentz force. One can easily imagine an Si-type sys¬
tem (where charge is a distinct unit) in which the Lorentz force takes the
form F = qE + ( q/c)\ x B, yielding the same dimensions for E and B.
A.6 Rationalized units
You might wonder why there are factors of 47r in the SI versions of
Coulomb’s law and the Biot-Savart law; see Eqs. (1.4) and (6.49). These
expressions would certainly look a bit less cluttered without these fac¬
tors. The reason is that the presence of 47 t’s in these laws leads to the
absence of such factors in Maxwell’s equations. And for various rea¬
sons it is deemed more important to have Maxwell’s equations be the
“clean” ones without the 47r factors. The procedure of inserting 47r into
Coulomb’s law and the Biot-Savart law, in order to keep them out of
Maxwell’s equations, is called “rationalizing” the units. Of course, for
768
Differences between SI and Gaussian units
people concerned more with applications of Coulomb’s law than with
Maxwell’s equations, this procedure might look like a step in the wrong
direction. But since Maxwell’s equations are the more fundamental equa¬
tions, there is logic in this convention.
It is easy to see why the presence of 4 n factors in Coulomb’s law
and the Biot-Savart law leads to the absence of 4 tv factors in Gauss’s
law and Ampere’s law, which are equivalent to two of Maxwell’s equa¬
tions (or actually one and a half; Ampere’s law is supplemented with
another term). In the case of Gauss’s law, the absence of the 47r basi¬
cally boils down to the area of a sphere being 4 tt r~ (see the derivation
in Section 1.10). In the case of Ampere’s law, the absence of the 4jt is a
consequence of the reasoning in Sections 6.3 and 6.4, which again boils
down to the area of a sphere being 4tt r 2 (because Eq. (6.44) was writ¬
ten down by analogy with Eq. (6.30)). Or more directly: the 1/47T in the
Biot-Savart law turns into a l/2:r in the field from an infinite straight
wire (see Eq. (6.6)), and this 2n is then canceled when we take the line
integral around a circle with circumference 2 nr.
If there were no factors of 47r in Coulomb’s law or the Biot-Savart
law, then there would be factors of 4 n in Maxwell’s equations. This is
exactly what happens in the Gaussian system, where the “curlB” and
“divE” Maxwell equations each involve a 47r; see Eq. (9.20). Note,
however, that one can easily imagine a Gaussian-type system (that is,
one where the pre-factor in Coulomb’s law is dimensionless) that has
factors of 4i r in Coulomb’s law and the Biot-Savart law, and none in
Maxwell’s equations. This is the case in a variation of Gaussian units
called Heaviside-Lorentz units.
We begin this appendix with the definitions of all of the derived SI units
relevant to electromagnetism (for example, the joule, ohm, etc.)- We then
list the units of all of the main quantities that appear in this book (basi¬
cally, anything that has earned the right to be labeled with its own letter).
In SI units the ampere is the fundamental unit involving charge. The
coulomb is a derived unit, being defined as one ampere-second. However,
since most people find it more natural to think in terms of charge than
current, we treat the coulomb as the fundamental unit in this appendix.
The ampere is then defined as one coulomb/second.
For each of the main quantities listed, we give the units in terms of
the fundamental units (m, kg, s, C, and occasionally K), and then also
in terms of other derived units in certain forms that come up often. For
example, the units of electric field are kgmC -1 s -2 , but they are also
newtons/coulomb and volts/meter.
The various derived units are as follows:
SI units of common
quantities
newton (N)
joule (J)
ampere (A)
volt (V)
kgm
s 2
kg m 2
newton-meter = —=—
s 2
coulomb C
second s
joule kg m 2
coulomb C s 2
770
SI units of common quantities
farad (F)
ohm (£2)
watt (W)
coulomb C 2 s 2
volt kg m 2
volt kg m 2
ampere C 2 s
joule kg m 2
second s 3
tesla (T)
newton kg
coulomb • meter/second C s
henry (H)
volt kg m 2
ampere / second C 2
The main quantities are listed by chapter.
Chapter 1
charge q: C
kg m 3 N m 2
k in Coulomb’s law: —
C 2 s 2 C 2
eo:
C 2 s 2
kg m 3
C 2 _ C _ F
N m 2 V m m
E field (force per charge):
flux <f> (E field times area):
charge density X, cr, p:
kg m _ N _ V
C s 2 C m
kg m 3 N m 2
c c c
m ’ m 2 ’ m 3
Chapter 2
potential (f> (energy per charge):
kg m 2
Cs 2
J
C
= V
dipole moment p: Cm
Chapter 3
C 2 s 2 C
capacitance C (charge per potential): - ~ = — = F
kg m 2 V
SI units of common quantities
771
Chapter 4
current I (charge per time): — = A
s
current density J (current per area):
conductivity a (current density per field):
resistivity p (field per current density):
resistance R (voltage per current):
power P (energy per time):
C-s 1
kg m 3 S3 m
kgm 3
C 2 s
kgm 2
C 2 s ~ A
kgm 2 J
-V- = - = w
S i s
= £2m
V
= - = a
Chapter 5
Chapter 6
m
speed of light c: —
s
kg
B field (force per charge-velocity): — = T
Mo :
kg m Tm
C 2 ~
kg m
vector potential A: = T m
C A
surface current density J (current per length): — = —
ms m
Chapter 7
electromotive force £:
kg m 2 J
flux <!> (B field times area):
inductance M, L:
Cs 2
kg m 2
Cs
= — = A ^ = V
C
= T m
kg m 2 Vs
C 2
= — = H
A
772
SI units of common quantities
Chapter 8
admittance Y (current per voltage):
1
frequency co: -
s
quality factor Q: 1 ( dimensionless )
phase <p\ 1 ( dimensionless)
C 2 s A 1
kg m 2 V £2
kgm 2 V
impedance Z (voltage per current): —=— = — =
C s A
Chapter 9
ka J W
power density S (power per area): —r- = —r— = —-
m 2 s m 2
Chapter 10
dielectric constant at: 1 ( dimensionless )
dipole momentp: Cm
kg m 2
torque N: —^— = N m
atomic polarizability a/4jteo: m
polarization density P: —=■
m-
electric susceptibility Xe■ 1 ( dimensionless )
C 2 s 2 C 2
permittivity e:
kg m 3 N m 2
C
displacement vector D: —-
m-
temperature T: K
kgm 2 J
Boltzmann’s constant k:
s 2 K K
SI units of common quantities
773
Chapter 11
magnetic moment m:
angular momentum L:
Planck’s constant Ir.
Cm 2
s
kgm 2
s
kgm 2
s
= Am 2
= J s
J
T
C A J
magnetization M (m per volume): — = — = - 7
m s m T m J
magnetic susceptibility 1 ( dimensionless )
C A
H field: — = -
ms m
kgm Tm
permeability /r: — 7 - = -
C z A
Unit conversions
In this appendix we list, and then derive, the main unit conversions
between the SI and Gaussian systems. As you will see below, many of the
conversions involve simple plug-and-chug calculations involving conver¬
sions that are already known. However, a few of them (charge, B field, H
field) require a little more thought, because the relevant quantities have
different definitions in the two systems.
C.1 Conversions
Except for the first five (nonelectrical) conversions below, we technic¬
ally shouldn’t be using “=” signs, because they suggest that the units in
the two systems are actually the same, up to a numerical factor. This is
not the case. All of the electrical relations involve charge in one way or
another, and a coulomb cannot be expressed in terms of an esu. This is
a consequence of the fact that the esu is defined in terms of the other
Gaussian units; see Appendix A for a discussion of how the coulomb
and esu differ. The proper way to express, say, the sixth relation below is
“1 coulomb is equivalent to 3 • 10 9 esu.” But well generally just use the
“=” sign, and you’ll know what we mean.
The “[3]” in the following relations stands for the “2.998” that
appears in the speed of light, c — 2.998 ■ 10 8 m/s. The coulomb-esu dis¬
cussion below explains how this arises.
time: 1 second = 1 second
length: 1 meter = 10 2 centimeter
1 kilogram = 10 3 gram
mass:
C.2 Derivations
775
force:
energy:
charge:
E potential:
E field:
capacitance:
resistance:
resistivity:
inductance:
B field:
El field:
1 newton:
1 joule:
1 coulomb:
1 volt =
1 volt/meter
1 farad
I ohm
1 ohm-meter
1 henry
1 tesla
1 amp/meter
I O' dyne
10 7 erg
[ 3 ] • 10 9 esu
1
[ 3 ]-
10 2 J
1
[ 3 ]-
10 4
[ 3] 2
■10 11
1
[ 3] 2
•10“
1
[ 3] 2
• 10 9
1
[ 3] 2
•10 11
10 4 .
gauss
An •
10 -3
s/cm
s-/cm
C.2 Derivations
C.2.1 Force: newton vs. dyne
1 newton = 1
kgm (1000g)( 100cm) 5 gem 5
—— — - y -= 10 —= 10' dynes.
(C.l)
C.2.2 Energy: joule vs. erg
kg m 2
1 joule = 1
(1000 g)(100cm) 2 7 gcm 2 in 7
- y -= 10' —= 10 ergs.
(C. 2 )
C.2.3 Charge: coulomb vs. esu
From Eqs. ( 1 . 1 ) and ( 1 . 2 ), two charges of 1 coulomb separated by a dis¬
tance of 1 m exert a force on each other equal to 8.988 • 10 9 N ~ 9 ■ 10 9 N,
or equivalently 9 ■ 10 14 dynes. How would someone working in Gaussian
units describe this situation? In Gaussian units. Coulomb’s law gives the
force simply as q 2 /r 2 . The separation is 100 cm, so if 1 coulomb equals N
esu (with N to be determined), the 9 • 10 14 dyne force between the charges
can be expressed as
9 ■ 10 14 dyne = ( N ^ N 2 = 9 ■ 10 18 => N — 3 ■ 10 9 .
y (100 cm) 2
(C. 3 )
776
Unit conversions
So 1 coulomb equals 3 • 10 9 esu. If we had used the more exact value of k
in Eq. (1.2), the “3” in this result would have been replaced by V8.988 =
2.998, which is precisely the 2.998 that appears in the speed of light,
c = 2.998 ■ 10 8 m/s. The reason for this is the following.
If you follow through the above derivation while keeping things in
terms of k = 1 /Aneo, you will see that the number 3 ■ 10 9 is actually
y/{k} ■ 10 5 • 10 4 , where we have put the braces around k to signify that it
is just the number 8.988 • 10 9 without the SI units. (The factors of 10 5 and
10 4 come from the conversions to dynes and centimeters, respectively.)
But we know from Eq. (6.8) that eo — 1 //toc 2 , so we have k = fiQC 2 /Ait.
Furthermore, the numerical value of /to is {/to} = 47r • 10 -7 , so the
numerical value of k is \k] = {c} 2 ■ 10 -7 . Therefore, the number N that
appears in Eq. (C.3) is really
N = y/{k] ■ 10 9 = yacF.lO-W = {c} • 10 = 2.998 • 10 9 s= [3] • 10 9 .
(C.4)
C.2.4 Potential: volt vs. statvolt
1 erg 1
I 10 7 erg
1 volt = 1 — =
C [3] • 10 9 esu [3] • 10 2 esu [3] • 10 2
statvolt.
C.2.5 Electric field: volt/meter vs. statvolt/centimeter
1
volt _ [3] -10 2
statvolt
1 statvolt
meter
100 cm
[3] • 10 4 cm
(C.5)
(C.6)
C.2.6 Capacitance: farad vs. centimeter
1 farad = 1 — =
V
[3]•10 9
esu
1
= [3] 2 • 10 11
[3]■ 10 2
statvolt
esu
statvolt
(C.7)
We can alternatively write these Gaussian units as centimeters. This is
true because 1 statvolt = 1 esu/cm (because the potential from a point
charge is q/r), so 1 esu/statvolt = 1 cm. We therefore have
1 farad = [3] 2 • 10 11 cm.
(C.8)
C.2.7 Resistance: ohm vs. second/centimeter
1
,. T , - jr statvolt ,
, U 1 V , V [3] • 10 2 1 S
A C/s [3]-10 9 esu/s [3] 2 - 10 11 esu/statvolt
Is n
~ [3] 2 • 10 11 cm’ (C ' 9
where we have used 1 esu/statvolt = 1 cm.
C.2 Derivations
111
C.2.8 Resistivity: ohm-meter vs. second
1 ohm-meter
I
[3] 2 ■ 10 11 cm
— ) (100cm)
I
[3] 2 ■ 10 9
C.2.9 Inductance: henry vs. second 2 /centimeter
1 henry = 1 1 — = 1 V - ^' 102
statvolt
A/s C/s 2 [3] • 10 9 esu/s 2
1
1
[3] 2 -10 11 esu/statvolt [3] 2 - 10 11 cm’
where we have used 1 esu/statvolt = 1 cm.
(C.10)
(C.ll)
C.2.10 Magnetic field B: tesla vs. gauss
Consider a setup in which a charge of 1 C travels at 1 m/s in a direction
perpendicular to a magnetic field with strength 1 tesla. Equation (6.1)
tells us that the force on the charge is 1 newton. Let us express this fact
in terms of the Gaussian force relation in Eq. (6.9), which involves a
factor of c. We know that 1N = 10 5 dyne and 1C = [3] • 10 9 esu. If
we let 1 tesla = A gauss, then the way that Eq. (6.9) describes the given
situation is
, [3]-10 9 esu / cm\
10 5 dyne = 1 J -- (100 — (Agauss). (C.12)
[3] • 10 IU cm/s V s /
Since 1 gauss equals 1 dyne/esu, all the units cancel (as they must), and
we end up with A= 10 4 , as desired. This is an exact result because the
two factors of [3] cancel.
C.2.11 Magnetic field H: ampere/meter vs. oersted
The H field is defined differently in the two systems (there is a /iq in the
SI definition), so we have to be careful. Consider a B field of 1 tesla in
vacuum. What H field does this B field correspond to in each system? In
the Gaussian system, B is 10 4 gauss. But in Gaussian units H — B in vac¬
uum, so H = 10 4 oersted, because an oersted and a gauss are equivalent
units. In the SI system we have (you should verify these units)
B 1 tesla 10 7 A
Yj _______ 13)
/xo 4 it ■ 10“ 7 kg m/C 2 4 n m
Since this is equivalent to 10 4 oersted, we arrive at 1 amp/meter= 47T • 10 -3
oersted. Going the other way, 1 oersted equals roughly 80 amp/meter.
SI and Gaussian
formulas
The following pages provide a list of all the main results in this book, in
both SI and Gaussian units. After looking at a few of the corresponding
formulas, you will discover that transforming from SI units to Gaussian
units involves one or more of the three types of conversions discussed
below.
Of course, even if a formula takes exactly the same form in the two
systems of units, it says two entirely different things. For example, the
formula relating force and electric field is the same in both systems: F =
q E. But in SI units this equation says that a charge of 1 coulomb placed
in an electric field of 1 volt/meter feels a force of 1 newton, whereas in
Gaussian units it says that a charge of 1 esu placed in an electric field
of 1 statvolt/centimeter feels a force of 1 dyne. When we say that two
formulas are the “same,” we mean that they look the same on the page,
even though the various letters mean different things in the two systems.
The three basic types of conversions from SI to Gaussian units are
given in Sections D.l to D.3. We then list the formulas in Section D.4 by
chapter.
D.l Eliminating and /*o
Our starting point in this book was Coulomb’s law in Eq. (1.4). The SI
expression for this law contains the factor 1 /4jteo, whereas the Gaus¬
sian expression has no factor (or rather just a 1). To convert from SI
units to Gaussian units, we therefore need to set 47reo= 1, or equiva¬
lently €o= 1/4tt (along with possibly some other changes, as we will
see below). That is, we need to erase all factors of 4jreo that appear, or
equivalently replace all eo’s with l/4jr’s. In many formulas this change
D.2 Changing B to B/c
779
is all that is needed. A few examples are: Gauss’s law, Eq. (1.31) in the
list in Section D.4; 1 the field due to a line or sheet, Eqs. (1.39) and (1.40);
the energy in an electric field, Eq. (1.53); and the capacitance of a sphere
or parallel plates, Eqs. (3.10) and (3.15).
A corollary of the eo —> 1 /4tt rule is the no^-An/c 2 rule. We
introduced no in Chapter 6 via the definition no = 1 /for' 2 , so if we
replace eo with 1 /4 jt, we must also replace no with 4 n/c 2 . An example
of this no 4?r /c 2 rule is the force between two current-carrying wires,
Eq. (6.15).
It is also possible to use these rules to convert formulas in the other
direction, from Gaussian units to SI units, although the process isn’t quite
as simple. The conversion must (at least for conversions where only eo
and no are relevant) involve multiplying by some power of An e q (or
equivalently 4 jt/ noc 2 ). And there is only one power that will make the
units of the resulting SI expression correct, because eo has units, namely
C 2 s 2 kg -1 m -3 . For example, the Gaussian expression for the field due
to a sheet of charge is 2 no in Eq. (1.40) in the list below, so the SI
expression must take the form of 2jrer(4jreo)". You can quickly show
that 2na {An€o)~ 1 = er/2eo has the correct units of electric field (it suf¬
fices to look at the power of any one of the four units: kg, m, s, C).
D.2 Changing B to Blc
If all quantities were defined in the same way in the two systems of units
(up to factors of 47reo and An/no c2 X then the above rules involving eo
and no would be sufficient for converting from SI units to Gaussian units.
But unfortunately certain quantities are defined differently in the two sys¬
tems, so we can’t convert from one system to the other without knowing
what these arbitrary definitions are.
The most notable example of differing definitions is the magnetic
field. In SI units the Lorentz force (or rather the magnetic part of it) is
F = q\ x B, while in Gaussian units it is F = {q/c)\ x B. To convert
from an SI formula to a Gaussian formula, we therefore need to replace
every B with a B/c (and likewise for the vector potential A). An example
of this is the B field from an infinite wire, Eq. (6.6). In SI units we have
B = /iof/2jrr. Applying our rules for no and B, the Gaussian B field is
obtained as follows:
which is the correct result. Other examples involving the B —>■ B/c rule
include Ampere’s law, Eqs. (6.19) and (6.25); the Lorentz transforma¬
tions, Eq. (6.76); and the energy in a magnetic field, Eq. (7.79).
1 The “double” equations in the list in Section D.4, where the SI and Gaussian formulas
are presented side by side, are labeled according to the equation number that the SI
formula has in the text.
780
SI and Gaussian formulas
D.3 Other definitional differences
The above two conversion procedures are sufficient for all formulas up to
and including Chapter 9. However, in Chapters 10 and 11 we encounter
a number of new quantities (/,,. D, H, etc.), and many of these quantities
are defined differently in the two systems of units,- mainly due to histor¬
ical reasons. For example, after using the €q —* 1 /4tt rule in Eq. (10.41),
we see that we need to replace Xe by 4 jtXe in going from SI to Gaussian
units. The Gaussian expression is then given by
P /4tt\ P P
Xe ~ ~^E > ~ ) e ^ Xe ~ (D ' 2)
which is correct. This Xe 4?r x e rule is consistent with Eq. (10.42).
Similarly, Eq. (10.63) shows that D is replaced by D/47T.
On the magnetic side of things, a few examples are the following.
Equation (11.9) shows that m (and hence M) is replaced by cm when
going from SI to Gaussian units (because m = 7a -* cm = 7a
m = /a/c, which is the correct Gaussian expression). Also, Eqs. (11.69)
and (11.70) show that H is replaced by (c/47t)H. Let’s check that
Eq. (11.68) is consistent with these rules. The SI expression for H is
converted to Gaussian as follows:
1
H = —B-M
Mo
- (cM)
H = B - 4ttM,
(D.3)
which is the correct Gaussian expression. Although it is possible to
remember all the different rules and then convert things at will, there
are so many differing definitions in Chapters 10 and 11 that it is prob¬
ably easiest to look up each formula as you need it. But for Chapters 1-9,
you can get a lot of mileage out of the first two rules above, namely (1)
cq -» 1/47T, /co —> 47 r/c 2 , and (2) B —*■ B/c.
DA The formulas
In the pages that follow, the SI formula is given first, followed by the
Gaussian equivalent.
2
The preceding case with B is simply another one of these differences, but we have
chosen to discuss it separately because the B field appears so much more often in this
book than other such quantities.
D.4 The formulas
781
Chapter 1
Coulomb’s law (1.4):
F _ 1 q\qir
47reo r 2
potential energy (1.9):
u _ 1 qiqi
4-jt r
electric field (1.20):
E - ' qt
47reo r 2
force and field (1.21):
F = qE
flux (1.26):
$ = J E ■ da
Gauss’s law (1.31):
[ E ■ da = —
J € 0
field due to line (1.39):
E r =
2neor
field due to sheet (1.40):
b \M
ii
AE across sheet (1.41):
<7 ,
AE = —n
eo
field near shell (1.42):
o
E r = ~
eo
E/(area) on sheet (1.49):
F 1 , ,
A = 2< Ei+Ej >
energy in E field (1.53):
G = y J E 2 dv
Chapter 2
electric potential (2.4):
(p — — J E • ds
field and potential (2.16):
E = -V</>
potential and density (2.18):
0 = f
J 47T e 0 r
potential energy (2.32):
U = \ J P<p dv
dipole potential (2.35):
q(. cos 9
<P=~ A -J
p = qi
F =
U =
qxqjf
r 2
q\q 2
r z
(same)
(same)
J E ■ da = 47 xq
2X
E r = —
r
E — 2tuj
AE = 47rern
E r = 4t rcr
(same)
U = — [E z dv
8j r J
(same)
(same)
(same)
ql cos 0
dipole moment (2.35):
(same)
782
SI and Gaussian formulas
dipole field (2.36):
qi . „
E= -=- (2 cos 9 r + sin@ 6)
4i reor-’ v
at
divergence theorem (2.49):
fa
>
II
cs
fa
(same)
E and p (2.52):
divE = —
eo
divE =
E and 0 (2.70):
divE = — V 2 0
(same)
0 and p (2.72):
V 2 0 = - —
eo
V 2 0 =
Stokes’ theorem (2.83):
j F • ds — [ curl F • da
Jc Js
(same)
Chapter 3
charge and capacitance (3.7):
G = C0
(same)
sphere C (3.10):
C —— Alt q.
C — a
parallel-plate C (3.15):
C= «A
s
A
C = -
4 TVS
energy in capacitor (3.29):
1 9
u = -cr
(same)
Chapter 4
current, current density (4.7):
I = J J • da
(same)
J and p (4.10):
d,vj = --
(same)
conductivity (4.11):
J = crE
(same)
Ohm’s law (4.12):
V = IR
(same)
resistivity (4.16):
j =G) e
(same)
resistance, resistivity (4.17):
pL
R= —
A
(same)
power (4.31):
P = IV = I 2 R
(same)
R, C time constant (4.43):
x — RC
(same)
D.4 The formulas
783
Chapter 5
Lorentz force (5.1):
F = ?E + fxB
F = qE + -y x B
c
charge in a region (5.2):
Q = e 0 J E • da
Q = — [ E • da
4t t J
E transformations (5.7):
E \\= E b e ± = Y e -L
(same)
E from moving Q (5.15):
E ' Q l ~f> 2
E ' Q i — p 2
Arte^r' 2 (1 — p 2 sin 2 d') 2 / 2
B 2 (1 — p 2 sin 2 O') 2 / 2
F transformations (5.17):
dp\\ _ d P\\ dp± _ 1 dp' ±
dt dt’ ’ dt y dt’
(same)
F from current (5.28):
F C/Vj
2 qv x I
V 27T ZqFC 2
~ 2
rc z
Chapter 6
B due to wire (6.3), ( 6 . 6 ):
speed of light ( 6 . 8 ):
Fona wire (6.14):
F between wires (6.15):
Ampere’s law (6.19):
D - I . Bol
B = Z- 7y — Z -
lire^rc- 2nr
B o^o
F — IBl
Boh hi
2jtr
/ B .* = bo I
rc
(no analog)
F =
IBl
c
F =
2I\l 2 l
f 47r
I B • ds = —I
(differential form) (6.25): curl B = /tqJ
curlB =
4?r
—J
c
vector potential (6.32): B = curl A
(same)
A and J (6.44):
Biot-Savart law (6.49):
B in solenoid (6.57):
A B across sheet (6.58):
F/(area) on sheet (6.63):
dB =
J dv
r
(j.qI dl x f
A = — f
4:r J
4:r
B z = fi 0 nl
A B = fj.Q J
F = (B+) 2 - (B-) 2
A 2/x 0
A = 1 [ ] -A
c J r
I dl x r
dB =-.—
Bz =
A B =
c r^
Annl
c
4t xj
F = Q B +) 2 - (B7) 2
A 87 r
784
SI and Gaussian formulas
E, B transforms (6.76):
E || = E ll
(same)
Bis = B II
(same)
E'j_ = y(E ± + P x cBx)
e 1l = y( E i. + P x Bj.)
cBj_ = y(cBx — P x Ej_)
Bj_ = Y (B± — ^ x Ej_)
-J x B
-J x B
Hall E t field (6.84):
E t = -
E, = -
nq
nqc
Chapter 7
electromotive force (7.5):
Faraday’s law (7.26):
(differential form) (7.31):
mutual inductance (7.37), (7.38):
self-inductance (7.57), (7.58):
L of toroid (7.62):
R, L time constant (7.69):
energy in inductor (7.74):
energy in B field (7.79):
£= - [ fds
q J
d<t>
£ = -
dt
3B
curl E-
dt
_ dl\
£ 2 \ = —M 21 —
dt
„ t//|
£ " =~ £| ,n
2tt
r =L//?
1
(/ = -LI 2
U =
-f
2/ro J
B 2 dv
(same)
1 d<&
c dt
curlE = —
1 3B
c dt
(same)
(same)
(same)
(same)
U = — [ B 2 dv
8tt J
Chapter 8
RLC time constant (8.8):
RLC frequency (8.9):
Q factor (8.12):
Iq for series RLC (8.38):
1
r = —
a
2 L
~R
Q — co-
h =
power
So
y/R 2 + (coL- 1/ft.C) 2
(same)
(same)
(same)
(same)
D.4 The formulas
785
<p for series RLC (8.39):
tan d> = -
RcoC R
(same)
resonant co (8.41):
1
<wo = ,—
VZc
(same)
width of I curve (8.45):
2| A<y| 1
COO Q
(same)
admittance (8.61):
I = YV
(same)
impedance (8.62):
v = zi
(same)
impedances (Table 8.1):
R, icoL, —i/toC
(same)
average power in R (8.81):
- V 2
D v rms
P ‘= R
(same)
average power (general) (8.85):
P — VxmJvms, COS (j)
(same)
Chapter 9
displacement current (9.15):
9E
Jd = fO^
1 9E
Jd T
4tt dt
Maxwell’s equations (9.17):
9B
curl E-
dt
1 9B
curl E-
c dt
9E
curl B = /j,oeo— + MoJ
dt
1 9E
curl B =-h
cdt
div E = —
eo
divE = 47rp
div B = 0
(same)
speed of wave (9.26), (9.27):
1
■s/Bo^o
V = c
E, B amplitudes (9.26), (9.27):
Bo
Eq = —= = cBq
VA'-oeo
Eo = Bo
power density (9.34):
S = €qE 2 C
E^c
Poynting vector (9.42):
B ExB
S c F V R
MO
An
invariant 1 (9.51):
E' B' = E B
(same)
invariant 2 (9.51):
E' 2 - c 2 B' 2 =E 2 - c 2 B 2
E' 2 - B' 2 =E 2 -
786
SI and Gaussian formulas
Chapter 10
dielectric constant (10.3):
dipole moment (10.13):
dipole potential (10.14):
dipole (E r ,Eg) (10.18):
torque on dipole (10.21):
force on dipole (10.26):
polarizability (10.29):
polarization density (10.31):
4> due to column (10.34):
surface density (10.35):
average field (10.37):
susceptibility (10.41):
Xe and k (10.42):
E in polar sphere (10.47):
permittivity (10.56):
P divergence (10.61):
displacement D ( 10.63):
D divergence (10.64):
D for linear ( 10.65):
Xe for weak£ (10.73):
k = Q/Qq
p = J r' p dv'
r p
<M r) =
4jreor 2
(2 cos 9, sin 0)
47reor 3
N = p x E
F x = p grad E
p = aE
P = pN
<P =
a — P
P
<E) =-
co
Xe= —
€o E
Xe = K ~ 1
P
E m —
3co
e = at6 0
di v P = Pbound
D = foE + P
div D = Pf ree
D = eE
Xe
Np 2
Co kT
J bound — „
at
(same)
(same)
r p
0(r) =
r z
-^(2 cos 0, sin 6)
r 3
(same)
(same)
(same)
(same)
/ 1
1 \
, / 1
1 \
- ■
- -
(p = P da -
- -
V/2
n)
V/2
n)
(same)
(E) = —4ttP
Xe= ~
E
P
E
k - 1
47T
4^-P
Xe = -
Em = ^
(no analog)
(same)
D = E + 4ttP
div D = 47rp fre(
D = a:E
Xe
Np 2
~kT
bound current J (10.74):
(same)
D.4 The formulas
787
speed of wave (10.83):
v V* 7
(same)
E, B amplitudes (10.83):
E 0 = ^ = v'Z?o
s/K
F Bo
Eo=~/=
y/K
Chapter 11
la
m = —
c
dipole moment (11.9):
m = /a
vector potential (11.10):
Mmxr
m x r
4 jt r 2
^ 2
r z
dipole ( B r ,Bg ) (11.15):
Mow .
- 7 (2 cos 6, sin 0)
4jrr J
m
-=-(2 cos 6 , sin^)
r 5
force on dipole (11.23):
F = V(m B)
(same)
orbital m for e (11.29):
m = -L
2 m e
T
m = -L
2 m e c
polarizability (11.41):
Am e 2 r 2
Am e 2 r 2
B 4 m e
B 4 m e c 2
torque on dipole (11.47):
N = m x B
(same)
polarization density (11.51):
m
M =
volume
(same)
susceptibility Xm (11-52):
B
M = Xm -
Mo
M = XmB
X P m for weak B (11.53):
IM)Nm 2
Nm 2
Xpm ~ kT
Xpm- kT
surface density J (11.55):
J = M
J = Me
volume density J (11.56):
J = curl M
J — c curl M
H field (11.68):
B
H =- M
Mo
H = B - 4?rM
curl of H (11.69):
curl H = Jfree
4 Tt
CUrl H = -Jf r ee
C
(integrated form) (11.70):
J H ■ dl — /free
f 4 tc
/ H dl=
Xm (accepted def.) (11.72):
M “ XmH
(same)
permeability (11.74):
M = Mo(l + Xm)
M = 1 +4nxm
B and H (11.74):
B = iiH
(same)
788
SI and Gaussian formulas
Appendix H
tangential Eg (H.3):
power (H.7):
Eg
Pad
qa sin 0
Att€qc 2 R
6neo c 3
Eg
Piad
qa sin 0
c 2 R
2 q 2 a 2
3 c 3
In 1983 the General Conference on Weights and Measures officially
redefined the meter as the distance that light travels in vacuum during
a time interval of 1/299,792,458 of a second. The second is defined in
terms of a certain atomic frequency in a way that does not concern us
here. The nine-digit integer was chosen to make the assigned value of c
agree with the most accurate measured value to well within the uncer¬
tainty in the latter. Henceforth the velocity of light is, by definition,
299,792,458 meters/second. An experiment in which the passage of a
light pulse from point A to point B is timed is to be regarded as a mea¬
surement of the distance from A to B, not a measurement of the speed of
light.
While this step has no immediate practical consequences, it does
bring a welcome simplification of the exact relations connecting various
electromagnetic units. As we learn in Chapter 9, Maxwell’s equations for
the vacuum fields, formulated in SI units, have a solution in the form of
a traveling wave with velocity c = (p ofo) -1 ^ 2 - The SI constant po has
always been defined exactly as 47r • 10 -7 kg m/C 2 , whereas the value of
6o has depended on the experimentally determined value of the speed of
light, any refinement of which called for adjustment of the value of eo-
But now €o acquires a permanent and perfectly precise value of its own,
through the requirement that
(/ro6o) -1//2 = 299,792,458 meters/second. (E.l)
Exact relations
among SI and
Gaussian units
In the Gaussian system no such question arises. Wherever c is
involved, it appears in plain view, and all other quantities are defined
exactly, beginning with the electrostatic unit of charge, the esu, whose
definition by Coulomb’s law involves no arbitrary factor.
790
Exact relations among SI and Gaussian units
With the adoption of Eq. (E.l) in consequence of the redefinition
of the meter, the relations among the units in the systems we have been
using can be stated with unlimited precision. These relations are listed at
the beginning of Appendix C for the principal quantities we deal with.
In the list the symbol [3] stands for the precise decimal 2.99792458.
The exact numbers are uninteresting and for our work quite unnec¬
essary. It is sheer luck that [3] happens to be so close to 3, an accidental
consequence of the length of the meter and the second. When 0.1 per¬
cent accuracy is good enough we need only remember that “300 volts is
a statvolt” and “3 ■ 10 9 esu is a coulomb.” Much less precisely, but still
within 12 percent, a capacitance of 1 cm is equivalent to 1 picofarad.
An important SI constant is (/xo/eo) 1//2 , which is a resistance in
ohms. Since eo = l//xoc 2 , this resistance equals /xoc. Using the exact
values of /xo and c, we find (po/eo) 1 ^ 2 = 40t r • [3] ohms 376.73 ohms.
One tends to remember it, and even refer to it, as “377 ohms.” It is the
ratio of the electric field strength E, in volts/meter, in a plane wave in vac¬
uum, to the strength, in amperes/meter, of the accompanying magnetic
field H. For this reason the constant (/xo/fo) 1/2 is sometimes denoted by
Zq and called, rather cryptically, the impedance of the vacuum. In a plane
wave in vacuum in which E rmii is the rms electric field in volts/meter, the
mean density of power transmitted, in watts/m 2 , is £ 2 ms /Z().
The logical relation of the SI electrical units to one another now
takes on a slightly different aspect. Before the redefinition of the meter,
it was customary to designate one of the electrical units as primary, in
this sense: its precise value could, at least in principle, be established by
a procedure involving the SI mechanical and metrical units only. Thus
the ampere, to which this role has usually been assigned, was defined in
terms of the force in newtons between parallel currents, using the relation
in Eq. (6.15). This was possible because the constant /xo in that relation
has the precise value 47r ■ 10 -7 kg m/C 2 . Then with the ampere as the
primary electrical unit, the coulomb was defined precisely as 1 ampere-
second. The coulomb itself, owing to the presence of eo in Coulomb’s
law, was not eligible to serve as the primary unit. Now with eo as well
as /xo assigned an exact numerical value, the system can be built up with
any unit as the starting point. All quantities are in this sense on an equal
footing, and the choice of a primary unit loses its significance. Never a
very interesting question anyway, it can now be relegated to history.
We begin this appendix by listing the main vector operators (gradient,
divergence, curl, Laplacian) in Cartesian, cylindrical, and spherical coor¬
dinates. We then talk a little about each operator - define things, derive
a few results, give some examples, etc. You will note that some of the
expressions below are rather scary looking. However, you won’t have to
use their full forms in this book. In the applications that come up, invari¬
ably only one or two of the terms in the expressions are nonzero.
Curvilinear
coordinates
F.1 Vector operators
F.1.1 Cartesian coordinates
ds = dx x
V = x-
dx
w 9/-
V/ = — x ■
dx
V ■ A =
9Aj
dx
t (d A-
V x A = —- -
dy y-
„ 9
4
dy
i a>/„
dy
9 Ay
“ir 4
dy
9A V
■ dz z,
„ 9
- z—,
3 z
. a
-Z,
dz
9A,
dz ’
\ dy
dx 2
dz
-J 3-/ 9 /
V“f = — H--
1 "- 9 9 y 2
9A£
9z
aV
9z 2
9A,\ „ / dA y dA x \„
dx ) ^ + \ dx dy )
(F.l)
792
Curvilinear coordinates
F.1.2 Cylindrical coordinates
ds = dr r + r d6 0 + dz z.
„ d
V = r-
dr
3/ ~
V/= fi-
dr
V • A = -
r
-i a A 3
0 -h Z —,
r dd dz
1 3/ a ^ 3/ .
- 0 H-z,
r 90 dz
3 (rA r )
13A e
3A-
dr
r 30 +
dz '
1 dA z
9 M f +
fdAr
r dd
dz ) +
l dz
Mz
dr
1 /d(rAff) dA r \
r V dr 30 / Z ’
1 3
r dr \ dr
3 f
1 3 2 /
30 2
32/
dz 2
(F.2)
F.1.3 Spherical coordinates
ds — drr + r d6 6 + r sin 0 d(f> 0,
1 3
V = r— -
dr
df „
V/= r
dr
~1 3
0 -1
r 30
1 3/ A
-- 0
r 30
0
rsin0 30 '
1 3/
0,
V • A =
1 3(r 2 A f .)
rsin0 30
1 3(Aesin0)
V x A =
1
rsin0 V
dr rsin0 30
/3(A^sin0) 3Ag
30
r + -
1 / 3(rA e ) 3A r \ >
c\ 3r 30 J
1 3
V Z /= --
! 3 /
1
r 2 dr \ dr) r 2 sin0 30
1 3A 0
rsin0 30
1 3A r
sin0 30
d{rAj,)
dr
■ a d f
sin0 —
30
1
d 2 f
r 2 sin- 0 30 2
0,
(F.3)
F.2 Gradient
The gradient produces a vector from a scalar. The gradient of a func¬
tion /, written as V/' or grad/, may be defined 1 as the vector with the
1 We used a different definition in Section 2.3, but we will show below that the two
definitions are equivalent.
F.2 Gradient
793
property that the change in / brought about by a small change ds in
position is
df=Vf- ds. (F.4)
The vector Vf depends on position; there is a different gradient vector
associated with each point in the parameter space.
You might wonder whether a vector that satisfies Eq. (F.4) actually
exists. We are claiming that if/ is a function of, say, three variables, then
at every point in space there exists a unique vector, V/, such that for any
small displacement ds from a given point, the change in/ equals V/ ■ ds.
It is not immediately obvious why a single vector gets the job done for all
possible displacements ds from a given point. But the existence of such
a vector can be demonstrated in two ways. First, we can explicitly con¬
struct V/; we will do this below in Eq. (F.5). Second, any (well-behaved)
function looks like a linear function up close, and for a linear function
a vector V/ satisfying Eq. (F.4) does indeed exist. We will explain why
in what follows. However, before addressing this issue, let us note an
important property of the gradient.
From the definition in Eq. (F.4), it immediately follows (as mentioned
in Section 2.3) that Vf points in the direction of steepest ascent of /.
This is true because we can write the dot product Vf ■ ds as |S/| |rfs| cos 0,
where 0 is the angle between the vector Vf and the vector ds. So for a
given length of the vector ds, this dot product is maximized when 0=0.
We therefore want the displacement ds to point in the direction of Vf, if
we want to produce the maximum change in /.
If we consider the more easily visualizable case of a function of
two variables, the function can be represented by a surface above the
xy plane. This surface is locally planar; that is, a sufficiently small bug
walking around on it would think it is a (generally tilted) flat plane. If we
look at the direction of steepest ascent in the local plane, and then project
this line onto the xy plane, the resulting line is the direction of Vf; see
Fig. 2.5. The function / is constant along the direction perpendicular to
Vf. The magnitude of Vf equals the change in / per unit distance in the
parameter space, in the direction of Vf. Equivalently, if we restrict the
parameter space to the one-dimensional line in the direction of steepest
ascent, then the gradient is simply the standard single-variable derivative
in that direction.
We could alternatively work “backwards” and define the gradient as
the vector that points in the direction (in the parameter space) of steep¬
est ascent, with its magnitude equal to the rate of change in that direc¬
tion. It then follows that the general change in /, for any displacement
ds in the parameter space, is given by Eq. (F.4). This is true because
the dot product picks out the component of ds along the direction of Vf.
This component causes a change in/, whereas the orthogonal component
does not.
794
Curvilinear coordinates
Figure F.l.
Only the component of ds in the direction of the
gradient causes a change in/.
Figure F.l shows how this works in the case of a function of two vari¬
ables. We have assumed for simplicity that the local plane representing
the surface of the function intersects the xy plane along the x axis. (We
can always translate and rotate the coordinate system so that this is true
at a given point.) The gradient then points in the y direction. The point
P shown lies in the direction straight up the plane from the given point.
The projection of this direction onto the xy plane lies along the gradient.
The point Q is associated with a ds interval that doesn’t lie along the gra¬
dient in the xy plane. This ds can be broken up into an interval along the
x axis, which causes no change in /, plus an interval in the y direction,
or equivalently the direction of the gradient, which causes the change in
/ up to the point Q.
The preceding two paragraphs explain why the vector V/ defined by
Eq. (F.4) does in fact exist; any well-behaved function is locally linear,
and a unique vector V/ at each point will get the job done in Eq. (F.4) if/
is linear. But as mentioned above, we can also demonstrate the existence
of such a vector by simply constructing it. Let’s calculate the gradient in
Cartesian coordinates, and then in spherical coordinates.
F.2.1 Cartesian gradient
In Cartesian coordinates, a general change in / for small displacements
can be written as df = (3 f/dx)dx + (3/ /dy)dy + ( df/dz)dz . This is just
the start of the Taylor series in three variables. The interval ds is simply
(dx, dy, dz), so if we want Vf ■ ds to be equal to df, we need
V/ =
(V df <?A
\3x’ dy’ dz)
df . 3 / . 3 / „
— xH-yH-z,
dx dy J dz
(F.5)
in agreement with the Vf expression in Eq. (F.l). In Section 2.3 we took
Eq. (F.5) as the definition of the gradient and then discussed its other
properties.
F.2.2 Spherical gradient
In spherical coordinates, a general change in / is given by df =
( df/dr)dr + (df/dO)dO + (df/d(f>)d(p. However, the interval ds takes
a more involved form compared with the Cartesian ds. It is
ds — (dr, r dd,r sin 9 df) = dr r + rd6 0 + r sin$ df (/>. (F.6)
If we want Vf ■ ds to be equal to df, then we need
V/ =
df 1 9f_ 1 df\
dr' r 30’ rsinO dtp)
— f + 0
dr r 3 0
rsint? 3/
(F.7)
in agreement with Eq. (F.3).
F.3 Divergence
795
We see that the extra factors (compared with the Cartesian case)
in the denominators of the gradient come from the coefficients of the
unit vectors in the expression for ds. Similarly, the form of the gradient
in cylindrical coordinates in Eq. (F.2) can be traced to the fact that the
interval ds equals drr + rd6 0 + dzi. Since the extra factors that appear
in ds show up in the denominators of the V-operator terms, and since
the V operator determines all of the other vector operators, we see that
every result in this appendix can be traced back to the form of ds in the
different coordinate systems. For example, the big scary expression listed
in Eq. (F.3) for the curl in spherical coordinates is a direct consequence
of the ds = dr r + r d6 0 + r sin 6 dtp <j> interval.
Note that the consideration of units tells us that there must be a factor
of r in the denominators in the df/d6 and df /d<p terms in the spherical
gradient, and in the df/d6 term in the cylindrical gradient.
F.3 Divergence
The divergence produces a scalar from a vector. The divergence of a
vector function was defined in Eq. (2.47) as the net flux out of a given
small volume, divided by the volume. In Section 2.10 we derived the
form of the divergence in Cartesian coordinates, and it turned out to be
the dot product of the V operator with a vector A, that is, V ■ A. We use
the same method here to derive the form in cylindrical coordinates. We
then give a second, more mechanical, derivation. A third derivation is
left for Exercise F.2.
F.3.1 Cylindrical divergence, first method
Consider the small volume that is generated by taking the region in the
r-6 plane shown in Fig. F.2 and sweeping it through a span of z values
from a particular z up to z + Az (the z axis points out of the page). Fet’s
first look at the flux of a vector field A through the two faces perpendic¬
ular to the z direction. As in Section 2.10, only the z component of A is
relevant to the flux through these faces. In the limit of a small volume, the
area of these faces is r A r A 6. The inward flux through the bottom face
equals A z (z) r Ar A6 , and the outward flux through the top face equals
A z (z+ A z) r A r A 6. We have suppressed the r and 0 arguments of A z for
simplicity, and we have chosen points at the midpoints of the faces, as in
Fig. 2.22. The net outward flux is therefore
Figure F.2.
A small region in the r-6 plane.
faces = A z (z + A") r Ar A6 — A z (z) r A r A 6
| C4-(z+ Az)-A.(z) ^
r Ar A 9 Az
BA Z
3 z
r Ar A9 Az.
(F.8)
796
Curvilinear coordinates
Upon dividing this net outward flux by the volume r Ar Ad Az, we obtain
dA-/dz, in agreement with the third term in V ■ A in Eq. (F.2). This was
exactly the same argument we used in Section 2.10. The z coordinate
in cylindrical coordinates is, after all, basically a Cartesian coordinate.
However, things get more interesting with the r coordinate.
Consider the flux through the two faces (represented by the curved
lines in Fig. F.2) that are perpendicular to the f direction. The key point
to realize is that the areas of these two faces are not equal. The upper
right one is larger. So the difference in flux through these faces depends
not only on the value of A r , but also on the area. The inward flux through
the lower left face equals A r (r)[r Ad Az], and the outward flux through
the upper right face equals A r (r + Ar)[(r + Ar) Ad Az]. As above, we
have suppressed the 6 and z arguments for simplicity, and we have chosen
points at the midpoints of the faces. The net outward flux is therefore
^ r faces —
Upon dividing this net outward flux by the volume r Ar Ad Az, we have
a leftover r in the denominator, so we obtain (1 /r)(d(rA r )/dr), in agree¬
ment with the first term in Eq. (F.2).
For the last two faces, the ones perpendicular to the 0 direction, we
don’t have to worry about different areas, so we quickly obtain
(r + A r)A r (r + Ar) Ad Az - rA r (r) Ad Az
(r + A r)A r (r + Ar) — rA, (r)'
d(rA r )
dr
Ar
Ar Ad Az.
Ar Ad Az
(F.9)
$0 faces — Ag(9 + Ad) Ar Az - Ae{6) Ar Az
'Ag(d + Ad) - Ag(ey
dAg
dd
Ad
Ar Ad Az.
Ar Ad Az
(F.10)
Upon dividing this net outward flux by the volume r Ar Ad Az, we again
have a leftover r in the denominator, so we obtain (l/r)(dAg/dd), in
agreement with the second term in Eq. (F.2).
If you like this sort of calculation, you can repeat this derivation
for the case of spherical coordinates. However, it’s actually not too hard
to derive the general form of the divergence for any set of coordinates;
see Exercise F.3. You can then check that this general formula reduces
properly for spherical coordinates.
F.3 Divergence
797
F.3.2 Cylindrical divergence, second method
Let’s determine the divergence in cylindrical coordinates by explicitly
calculating the dot product,
/j -la A a \ /„ - A \
VA = ( r 8t +e ;^ +z fe)( r ' 4 ' +M * +zA =)- (FJ1)
At first glance, it appears that V ■ A doesn’t produce the form of the diver¬
gence given in Eq. (F.2). The second two terms work out, but it seems
like the first term should simply be 9A,./9r instead of (1 /r)(d(rA r )/dr).
However, the dot product does indeed correctly yield the latter term,
because we must remember that, in contrast with Cartesian coordinates,
in cylindrical coordinates the unit vectors themselves depend on position.
This means that in Eq. (F.ll) the derivatives in the V operator also act on
the unit vectors in A. This issue doesn’t come up in Cartesian coordinates
because x, y, and z are fixed vectors, but that is more the exception than
the rule. Writing A in the abbreviated form (A r ,Ag,A z ) tends to hide
important information. The full expression for A is i'A + OAg + zA z .
There are six quantities here (three vectors and three components), and
if any of these quantities vary with the coordinates, then these varia¬
tions cause A to change. The derivatives of the unit vectors that are
nonzero are
or ou
— = 0 and — = -r. (F.12)
90 90
To demonstrate these relations, we can look at what happens to r and
0 if we rotate them through an angle dO. Since the unit vectors have
length 1, we see from Fig. F.3 that r picks up a component of length
dO in the 0 direction, and 0 picks up a component of length d6 in the
—r direction. The other seven of the nine possible derivatives are zero
because none of the unit vectors depends on r or z, and furthermore z
doesn’t depend on 0.
Due to the orthogonality of the unit vectors, we quickly see that, in
addition to the three “corresponding” terms that survive in Eq. (F.ll),
one more term is nonzero:
■j 1 9
~ 1 /9r
r 90
, 9A,
0 - (?A,.) = 0 ■ - — A r + r —- ) = 0 ■ - 0A
90
90
A r
0 = —.
r
(F.13)
Equation (F.ll) therefore becomes
Figure F.3.
How the r and 6 unit vectors depend on 9.
9A,. 1 dAg 9A- A,.
—- H-- H-- + —.
dr r 90 dz r
(F.14)
The sum of the first and last terms here can be rewritten as the first term
in V • A in Eq. (F.2), as desired.
798
Curvilinear coordinates
F.4 Curl
The curl produces a vector from a vector. The curl of a vector function
was defined in Eq. (2.80) as the net circulation around a given small
area, divided by the area. (The three possible orientations of the area
yield the three components.) In Section 2.16 we derived the form of the
curl in Cartesian coordinates, and it turned out to be the cross product
of the V operator with the vector A, that is, V x A. Well use the same
method here to derive the form in cylindrical coordinates, after which
we derive it a second way, analogous to the above second method for the
divergence. Actually, we’ll calculate just the z component; this should
make the procedure clear. As an exercise you can calculate the other two
components.
F.4.1 Cylindrical curl, first method
The z component of V x A is found by looking at the circulation around
a small area in the r-6 plane (or more generally, in some plane parallel
to the r-6 plane). Consider the upper right and lower left (curved) edges
in Fig. F.2. Following the strategy in Section 2.16, the counterclockwise
line integral along the upper right edge equals Ag (r + Ar)[(r + A r) A 6],
and the counterclockwise line integral along the lower left edge equals
—Ag(r)[r At?]. We have suppressed the 6 and z arguments for simplicity,
and we have chosen points at the midpoints of the edges. Note that we
have correctly incorporated the fact that the upper right edge is longer
than the lower left edge (the same issue that came up in the above calcu¬
lation of the divergence). The net circulation along these two edges is
Cg sides = (r + Ar)Ag(r + A r) A 6 - rAg(r) A 6
(F.15)
Upon dividing this circulation by the area r A r Ad, we have a leftover r
in the denominator, so we obtain {\/r)(d{rAg)/dr), in agreement with
the first of the two terms in the z component of V x A in Eq. (F.2).
Now consider the upper left and lower right (straight) edges. The
counterclockwise line integral along the upper left edge equals —A r {6 +
A6) Ar, and the counterclockwise line integral along the lower right
edge equals A r (9) Ar. The net circulation along these two edges is
C r sides — — A r (0 + Ad) Ar + A r (d) Ar
dA r
~86
Ar Ad.
(F.16)
F.5 Laplacian
799
Upon dividing this circulation by the area r Ar AO, we again have a left¬
over r in the denominator, so we obtain — (l/r)(9A r /96), in agreement
with Eq. (F.2).
F.4.2 Cylindrical curl, second method
Our goal is to calculate the cross product.
V x A =
+ 0
13 „ 3 \
-h z— )
r96 dzj
(F.17)
while remembering that some of the unit vectors depend on the coor¬
dinates according to Eq. (F.12). As above, we’ll look at just the z com¬
ponent. This component arises from terms of the form r x 6 or 6 x r .
In addition to the two obvious terms of this form, we also have the one
involving § x (96/dO), which from Eq. (F.12) equals 6 x (—f) = z. The
complete z component of the cross product is therefore
(V x A), = r x
3 (OAg)
dr
+ 0 x
1 3(rA,.)
~r 96
+ 6 x
1 3 (9 As)
r 96
(dAg _ 13A, Ag\
\ 9r r 99 r )
(F.18)
The sum of the first and last terms here can be rewritten as the first term
in the z component of V x A in Eq. (F.2), as desired.
F.5 Laplacian
The Laplacian produces a scalar from a scalar. The Laplacian of a func¬
tion/ (written as V jf or V• V/) is defined as the divergence of the gradient
off. Its physical significance is that it gives a measure of how the average
value off over the surface of a sphere compares with the value off at
the center of the sphere. Let’s be quantitative about this.
Consider the average value of a function / over the surface of a
sphere of radius r. Call it / av g,r- If we choose the origin of our coordi¬
nate system to be the center of the sphere, then / av g,r can be written as
(with A being the area of the sphere)
r,.,, = = fSdQ = Aj/ da , (F.19)
where d£2 = sin 9 d6 d(p is the solid-angle element. We are able to
take the r 2 outside the integral and cancel it because r is constant over
the sphere. This expression for / avg ,r is no surprise, of course, because
the integral of dfl over the whole sphere is 4 tt. But let us now take the
d/dr derivative of both sides of Eq. (F.19), which will alow us to invoke
800
Curvilinear coordinates
the divergence theorem. On the right-hand side, the integration doesn’t
involve r, so we can bring the derivative inside the integral. This yields
(using f • r = 1)
df ivg ,r
dr
1
47T
/
r— • r d£l =
dr
1
4 nr 2
f
r— ■ rriffi.
dr
(F.20)
(Again, we are able to bring the r 2 inside the integral because r is con¬
stant over the sphere.) But f r 2 d£Z is just the vector area element of the
sphere, da. And f (df /dr) is the f component of V/’ in spherical coordi¬
nates. The other components of Vf give zero when dotted with da, so we
can write
dfavg.r
dr
(F.21)
The divergence theorem turns this into
dfdvg.r
dr
(F.22)
There are two useful corollaries of this result. First, if V jf = 0 every¬
where, then dfdvg.rjdr = 0 for all r. In other words, the average value
of / over the surface of a sphere doesn’t change as the sphere grows
(while keeping the same center). So all spheres centered at a given point
have the same average value off. In particular, they have the same aver¬
age value that an infinitesimal sphere has. But the average value over
an infinitesimal sphere is simply the value at the center. Therefore, if
= 0, then the average value off over the surface of a sphere (of any
size) equals the value at the center:
V2f = 0 => / avg , r =/center- (F.23)
This is the result we introduced in Section 2.12 and proved for the special
case of the electrostatic potential ([>.
Second, we can derive an expression for how / changes, for small
values of r. Up to this point, all of our results have been exact. We will
now work in the small-r approximation. In this limit we can say that V/
is essentially constant throughout the interior of the sphere (assuming
that / is well-enough behaved). So its value everywhere is essentially
the value at the center. The volume integral in Eq. (F.22) then equals
(47T r 3 /3)(V jO C ente r > and we have
dfi\gj
dr
1 4jrr 3
4nr 2 3
( V jO center
df ivg ,r
dr
^ (v ^° center* (F.24)
Exercises
801
Since (V^Ocenter is a constant, we can quickly integrate both sides of this
relation to obtain
r~
favg.r — /center H center (for small r), (F.25)
where the constant of integration has been chosen to give equality at
r = 0. We see that the average value off over a (small) sphere grows
quadratic ally, with the quadratic coefficient being 1 /6 times the value of
the Laplacian at the center.
Let’s check this result for the function /'(r, 9, <p) = r 2 , or equivalently
f(x,y, z ) = x 2 +y 2 + z 2 ■ By using either Eq. (F.l) or Eq. (F.3) we obtain
V^f = 6. If our sphere is centered at the origin, then Eq. (F.25) gives
/avg.r = 0 + (r 2 /6)(6) = r 2 , which is correct because / takes on the
constant value of r 2 over the sphere. In this simple case, the result is
exact for all r.
F.5.1 Cylindrical Laplacian
Let’s explicitly calculate the Laplacian in cylindrical coordinates by cal¬
culating the divergence of the gradient off. As we’ve seen in a few cases
above, we must be careful to take into account the position dependence
of some of the unit vectors. We have
V-V/ =
+ 0
13 „ 9 \
-4~ z — ]
r dd dzJ
* i a f
0 ——
r d9
. 3 /
'dz
(F.26)
In addition to the three “corresponding” terms, we also have the term
involving 0 ■ (dr/dO), which from Eq. (F.12) equals 0 ■ 0 = 1. So this
fourth term reduces to (l/r)(9//3r). The Laplacian is therefore
J dr\dr) r dO \r dO J dz\dzj r dr
(F.27)
_ d 2 f 1 3 2 / d 2 f 1 3 f
dr 2 + r 2 dO 2 + dz 2 r dr
The sum of the first and last terms here can be rewritten as the first term
in the V^f expression in Eq. (F.2), as desired.
Exercises
F.l Divergence using two systems **
(a) The vector A = xx + yy in Cartesian coordinates equals the
vector A = rr in cylindrical coordinates. Calculate V ■ A in
both Cartesian and cylindrical coordinates, and verify that the
results are equal.
(b) Repeat (a) for the vector A = xx + 2yy. You will need to find
the cylindrical components of A, which you can do by using
x = r cos 0 — 0 sin 9 and y = f sin 0 + 0 cos 9. Alternatively,
802
Curvilinear coordinates
you can project A onto the unit vectors, f = x cos 9 + y sin 0
and 6 — —x sin0 + y cos 9.
F.2 Cylindrical divergence ***
Calculate the divergence in cylindrical coordinates in the follow¬
ing way. We know that the divergence in Cartesian coordinates is
V ■ A = dA x /dx + dAy/dy + 3 A z /dz. To rewrite this in terms of
cylindrical coordinates, show that the Cartesian derivative opera¬
tors can be written as (the 3/3 z derivative stays the same)
3
3x
3
3 y
= cos 9 -
dr
sin 0
1 3
7 30 ’
. „ 3 13
= sintl-f- cos 0 -,
dr r 3 9
(F.28)
and that the components of A can be written as (A, stays the same)
A x — A r cos 9 — Aq sin 9,
A y = A r sin 9 + Aq cos 9. (F.29)
Then explicitly calculate V • A = dA x /dx + dA y /dy + dA z /dz. It
gets to be a big mess, but it simplifies in the end.
F.3 General expression for divergence ***
Let xi, X 2 , X 3 be the (not necessarily Cartesian) basis vectors of
a coordinate system. For example, in spherical coordinates these
vectors are f, 6, </>. Note that the ds line elements listed at the
beginning of this appendix all take the form of
ds = f\ dx i xi +/ 2 dx 2 x 2 +/3 dx 3 x 3 , (F.30)
where the / factors are (possibly trivial) functions of the coordi¬
nates. For example, in Cartesian coordinates, f\ ,,/ 2 ,./3 are 1,1,1;
in cylindrical coordinates they are 1 , r, 1 ; and in spherical coor¬
dinates they are 1, r, rsinfh As we saw in Section F.2, these val¬
ues off determine the form of V (the / factors simply end up in
the denominators), so they determine everything about the various
vector operators. Show, by applying the first method we used in
Section F.3, that the general expression for the divergence is
V • A =
1 r WaA,)
flfzfi
3xi
3(/l/ 3 A 2 )
3 x 2
diflfiAs)
8 x 3
(F.31)
Verify that this gives the correct result in the case of spherical
coordinates. (The general expression for the curl can be found in a
similar way.)
Exercises
803
F.4 Laplacian using two systems **
(a) The function/ = x 2 + y 1 in Cartesian coordinates equals the
function/ = r 2 in cylindrical coordinates. Calculate V/ in
both Cartesian and cylindrical coordinates, and verify that the
results are equal.
(b) Repeat (a) for the function/ = x 4 + y 4 . You will need to
determine what/ looks like in cylindrical coordinates.
F.5 “Sphere" averages in one and two dimensions **
Equation (F.25) holds for a function / in 3D space, but analogous
results also hold in 2D space (where the “sphere” is a circle bound¬
ing a disk) and in ID space (where the “sphere” is two points
bounding a line segment). Derive those results. Although it is pos¬
sible to be a little more economical in the calculations by stripping
off some dimensions at the start, derive the results in a 3D manner
exactly analogous to the way we derived Eq. (F.25). For the 2D
case, the relevant volume is a cylinder, with / having no depen¬
dence on z■ For the ID case, the relevant volume is a rectangular
slab, with / having no dependence on y or z- The ID result should
look familiar from the standard ID Taylor series.
F .6 Average over a cube ***
By using the second-order Taylor expansion for a function of three
Cartesian coordinates, show that the average value of a function
/ over the surface of a cube of side 21 (with edges parallel to the
coordinate axes) is
5 1 2 ,
/avg —/center H 7<r C^jOcenter- (F.32)
io
You should convince yourself why the factor of 5/18 here is cor¬
rectly larger than the 1/6 in Eq. (F.25) and smaller than (V3) 2 /6.
A short review of
special relativity
G.1 Foundations of relativity
We assume that the reader has already been introduced to special
relativity. Here we shall review the principal ideas and formulas that are
used in the text beginning in Chapter 5. Most essential is the concept of
an inertial frame of reference for space-time events and the transforma¬
tion of the coordinates of an event from one inertial frame to another.
A frame of reference is a coordinate system laid out with measuring
rods and provided with clocks. Clocks are everywhere. When something
happens at a certain place, the time of its occurrence is read from a clock
that was at, and stays at, that place. That is, time is measured by a local
clock that is stationary in the frame. The clocks belonging to the frame
are all synchronized. One way to accomplish this (not the only way) was
described by Einstein in his great paper of 1905. Light signals are used.
From a point A, at time t a, a short pulse of light is sent out toward a
remote point B. It arrives at B at the time f#, as read on a clock at B,
and is immediately reflected back toward A, where it arrives at t' A . If
tg = (f/t + t' A )/ 2, the clocks at A and B are synchronized. If not, one
of them requires adjustment. In this way, all clocks in the frame can be
synchronized. Note that the job of observers in this procedure is merely
to record local clock readings for subsequent comparison.
An event is located in space and time by its coordinates x, y, z, t
in some chosen reference frame. The event might be the passage of a
particle at time t \, through the space point (xi,yi,zi). The history of
the particle’s motion is a sequence of such events. Suppose the sequence
has the special property that x = v x t, y — v y t, z = v z t, at every time f,
with v x , v y , and v, constant. That describes motion in a straight line at
G.2 Lorentz transformations
805
constant speed with respect to this frame. An inertial frame of reference
is a frame in which an isolated body, free from external influences, moves
in this way. An inertial frame, in other words, is one in which Newton’s
first law is obeyed. Behind all of this, including the synchronization of
clocks, are two assumptions about empty space: it is homogeneous (that
is, all locations in space are equivalent) and it is isotropic (that is, all
directions in space are equivalent).
Two frames, let us call them F and F', can differ in several ways.
One can simply be displaced with respect to the other, the origin of coor¬
dinates in F' being fixed at a point in F that is not at the F coordinate
origin. Or the axes in F' might not be parallel to the axes in F. As for the
timing of events, if F and F 1 are not moving with respect to one another,
a clock stationary in F is stationary also in /• '. In that case we can set all
F' clocks to agree with the F clocks and then ignore the distinction. Dif¬
ferences in frame location and frame orientation only have no interesting
consequences if space is homogeneous and isotropic. Suppose now that
the origin of frame F' is moving relative to the origin of frame F. The
description of a sequence of events by coordinate values and clock times
in F can differ from the description of the same events by space coordi¬
nate values in F' and times measured by clocks in F'. How must the two
descriptions be related? In answering that we shall be concerned only
with the case in which F is an inertial frame and F' is a frame that is
moving relative to F at constant velocity and without rotating. In that
case F' is also an inertial frame.
Special relativity is based on the postulate that physical phenomena
observed in different inertial frames of reference appear to obey exactly
the same laws. In that respect one frame is as good as another; no frame
is unique. If true, this relativity postulate is enough to determine the
way a description of events in one frame is related to the description
in a different frame of the same events. In that relation there appears a
universal speed, the same in all frames, whose value must be found by
experiment. Sometimes added as a second postulate is the statement that
a measurement of the velocity of light in any frame of reference gives the
same result whether the light’s source is stationary in that frame or not.
One may regard this as a statement about the nature of light rather than
an independent postulate. It asserts that electromagnetic waves in fact
travel with the limiting speed implied by the relativity postulate. The
deductions from the relativity postulate, expressed in the formulas of
special relativity, have been precisely verified by countless experiments.
Nothing in physics rests on a firmer foundation.
G.2 Lorentz transformations
Consider two events, A and B, observed in an inertial frame
F. Observed, in this usage, is short for “whose space-time coordinates are
determined with the measuring rods and clocks of frame F.” (Remember
806
A short review of special relativity
that our observers are equipped merely with pencil and paper, and we
must post an observer at the location of every event!) The displacement
of one event from the other is given by the four numbers
x B - x A , y B - y A , z B - z A , t B - t A . (G.l)
The same two events could have been located by giving their coordi¬
nates in some other frame F'. Suppose F' is moving with respect to F in
the manner indicated in Fig. G.l. The spatial axes of F' remain parallel
to those in F, while, as seen from F, the frame F' moves with speed v in
the positive x direction. This is a special case, obviously, but it contains
most of the interesting physics.
Event A, as observed in F', occurred at x' A ,y' A ,z! A , f A , the last of these
numbers being the reading of a clock belonging to (that is, stationary in)
F'. The space-time displacement, or interval between events A and B in
F', is not the same as in F. Its components are related to those in F by
the Lorentz transformation,
X ' B - x 'a = Y( x b ~ x a ) - Pyc(t B - 4),
y' B -y'A = ye - yA,
Zft — Zb Za 1
t' B - t A = y(t B - 4) - fy(x B - x A )/c. (G.2)
In these equations c is the speed of light, f = v/c, and y =
1 /yj 1 — f 2 . The inverse transformation has a similar appearance - as
it should if no frame is unique. It can be obtained from Eq. (G.2) sim¬
ply by exchanging primed and unprimed symbols and reversing the sign
of ft, as you can verify by explicitly solving for the quantities x B — x A
and 4 — 4 .
Two events A and B are simultaneous in F if t H — t A = 0. But that
does not make t' B — t' A = 0 unless x B = x A . Thus events that are simul¬
taneous in one inertial frame may not be so in another. Do not confuse
this fundamental “relativity of simultaneity” with the obvious fact that
an observer not equally distant from two simultaneous explosions will
receive light flashes from them at different times. The times t' A and t' B are
recorded by local clocks at each event, clocks stationary in F' that have
previously been perfectly synchronized.
Consider a rod stationary in F' that is parallel to the x! axis and
extends from x A to x B . Its length in F' is just x B — x A . The rod’s length as
measured in frame F is the distance x B — x A between two points in the
frame F that its ends pass simultaneously according to clocks in F. For
these two events, then, t B — 4 = 0. With this condition the first of the
Lorentz transformation equations above gives us at once
x B -x A = {x B -x A )/y.
(G.3)
G.2 Lorentz transformations
807
Figure G.l.
Two frames moving with relative speed v. The
“E” is stationary in frame F. The “IT is stationary
in frame F'. In this example p = v/c = 0.866;
y = 2. (a) Where everything was, as determined
by observers in F at a particular instant of time t
according to clocks in F. (b) Where everything
was, as determined by observers in F' at a
particular instant of time t' according to
clocks in F'.
Question: Suppose the clocks in the two frames
happened to be set so that the left edge of the E
touched the left edge of the L at t = 0 according
to a local clock in F, and at 7 = 0 according to a
local clock in F'. Let the distances be in feet and
take c as 1 foot/nanosecond. What is the
reading t of all the F clocks in (a)? What is the
reading / of all the F' clocks in (b)?
Answer: t = 4.62 nanoseconds; / = 4.04
nanoseconds. If you don't agree, study the
example again.
This is the famous Lorentz contraction. Loosely stated, lengths between
fixed points in F', if parallel to the relative velocity of the frames, are
judged by observers in F to be shorter by the factor 1 jy. This statement
remains true if F' and F are interchanged. Lengths perpendicular to the
relative velocity measure the same in the two frames.
808
A short review of special relativity
Consider one of the clocks in F'. It is moving with speed v through
the frame F. Let us record as t' A its reading as it passes one of our local
clocks in F\ the local clock reads at that moment t A . Later this moving
clock passes another F clock. At that event the local F clock reads t B ,
and the reading of the moving clock is recorded as t' B . The two events are
separated in the F frame by a distance x B — x A = v(t B — 1 A ). Substituting
this into the fourth equation of the Lorentz transformation, Eq. (G.2), we
obtain
t' B — t' A — y(t B — t A ){\ — f}~) — (f fl — t A )/y. (G.4)
According to the moving clock, less time has elapsed between the two
events than is indicated by the stationary clocks in F. This is the time
dilation that figures in the “twin paradox.” It has been verified in many
experiments, including one in which an atomic clock was flown around
the world.
Remembering that “moving clocks run slow, by the factor 1 /y,” and
that “moving graph paper is shortened parallel to its motion by the factor
1 /y" you can often figure out the consequences of a Lorentz transforma¬
tion without writing out the equations. This behavior, it must be empha¬
sized, is not a peculiar physical property of our clocks and paper, but is
intrinsic in space and time measurement under the relativity postulate.
G.3 Velocity addition
The formula for the addition of velocities, which we use in Chapter 5,
is easily derived from the Lorentz transformation equations. Suppose an
object is moving in the positive x direction in frame F with velocity u x .
What is its velocity in the frame F'l To simplify matters let the moving
object pass the origin at f = 0. Then its position in F at any time t is
simply x = u x t. To simplify further, let the space and time origins of F
and F' coincide. Then the first and last of the Lorentz transformation
equations become
x' = yx — fiyct and t'= yt — fiyx/c. (G.5)
By substituting u x t for x on the right side of each equation, and dividing
the first by the second, we get
x' u x — P c
t' 1 — Pu x /c
(G.6)
On the left we have the velocity of the object in the F' frame, u' x . The
formula is usually written with v instead of fic.
u x ~ v
1 — u x v/c 2
u
(G.7)
G.4 Energy, momentum, force
809
By solving Eq. (G.7) for u x you can verify that the inverse is
u' x + v
u x = ---x ,
1 + u' x v/c 2
(G.8)
and that in no case will these relations lead to a velocity, either u x or
u x , larger than c. As with the inverse Lorentz transformation, you can
also obtain Eq. (G.8) from Eq. (G.7) simply by exchanging primed and
unprimed symbols and reversing the sign of v.
A velocity component perpendicular to v, the relative velocity of the
frames, transforms differently, of course. Analogous to Eq. (G.5), the
second and last of the Lorentz transformation equations are
/ —y and t' = yt — fiyx/c. (G.9)
If we have x = u x t and y — u y t in frame F (in general the object can
be moving diagonally), then we can substitute these into Eq. (G.9) and
divide the first equation by the second to obtain
y' m v , M v
- =- z - => u' = - z -z- . (G.10)
t' y(l — fiu x /c) y y(l — u x v/c 2 )
In the special case where u x = 0 (which means that the velocity points
in the y direction in frame F), we have u' y = u y /y. That is, the y speed
is slower in the frame F' where the object is flying by diagonally. In the
special case where u x = v (which means that the object travels along
with the F' frame, as far as the x direction is concerned), you can show
that Eq. (G.10) reduces to u' = yu y =>• u y = u' v /y. This makes sense;
the object has u' x = 0, so this result is analogous to the preceding u' =
u y /y result for the u x = 0 case. In effect we have simply switched the
primed and unprimed labels. These special cases can also be derived
directly from time dilation.
G.4 Energy, momentum, force
A dynamical consequence of special relativity can be stated as follows.
Consider a particle moving with velocity u in an inertial frame F. We
find that energy and momentum are conserved in the interactions of this
particle with others if we attribute to the particle a momentum and an
energy given by
p = yniQU and E = ymoc 2 , (G.ll)
where ihq is a constant characteristic of that particle. We call m o the rest
mass (or just the mass ) of the particle. It could have been determined
in a frame in which the particle is moving so slowly that Newtonian
mechanics applies - for instance, by bouncing the particle against some
standard mass. The factor y multiplying m o is (1 — m 2 /c 2 ) -1 ' 2 , where u
is the speed of the particle as observed in our frame F.
810
A short review of special relativity
Given p and E, the momentum and energy of a particle as observed
in F, what is the momentum of that particle, and its energy, as observed
in another frame F'l As before, we assume F' is moving in the positive x
direction, with speed v, as seen from F. The transformation turns out to
be this:
p'x - YPx - PyE/c,
Py = Py >
p'z = Pz ’
E' = yE- Pycp x . (G.12)
Note that f>c is here the relative velocity of the two frames, as it was in
Eq. (G.2), not the particle velocity.
Compare this transformation with Eq. (G.2). The resemblance would
be perfect if we considered cp instead of p in Eq. (G.12), and ct rather
than t in Eq. (G.2). A set of four quantities that transform in this way is
called a four-vector.
The meaning of force is rate of change of momentum. The force
acting on an object is simply dp/dt, where p is the object’s momentum in
the chosen frame of reference and t is measured by clocks in that frame.
To find how forces transform, consider a particle of mass mo initially at
rest at the origin in frame F upon which a force / acts for a short time
At. We want to find the rate of change of momentum dp'/dt' , observed
in a frame F'. As before, we shall let F' move in the x direction as seen
from F. Consider first the effect of the force component f x . In time At,
p x will increase from zero to/) Af, while x increases by
Ax=I(^W (G.13)
2 V mo )
and the particle’s energy increases by A E = (f x At) 2 /2mo; this is the
kinetic energy it acquires, as observed in F. (The particle’s speed in F is
still so slight that Newtonian mechanics applies there.) Using the first of
Eqs. (G.12) we find the change in p' x \
A p'x = yApx - PyAE/c, (G.14)
and using the fourth of Eqs. (G.2) gives
At' = y At — PyAx/c. (G.15)
Now both A E and Ax are proportional to (At) 2 , so when we take the
limit At —*■ 0, the last term in each of these equations will drop out,
giving
dp’ x = Hm A p' x = y(f x At)
dt' Ar'->o At' y At
(G.16)
G.4 Energy, momentum, force
811
Conclusion: the force component parallel to the relative frame
motion has the same value in the moving frame as in the rest frame of the
particle.
A transverse force component behaves differently. In frame F, A p y =
f y At. But now A p' y = Ap v , and Af = y At, so we get
dp' f y At f v
-AA = — = J A . (G.17)
dt’ y At y
A force component perpendicular to the relative frame motion, observed
in F', is smaller by the factor 1 fy than the value determined by observers
in the rest frame of the particle.
The transformation of a force from F' to some other moving frame
F" would be a little more complicated. We can always work it out, if we
have to, by transforming to the rest frame of the particle and then back to
the other moving frame.
We conclude our review with a remark about Lorentz invariance. If
you square both sides of Eq. (G.12) and remember that y 2 — f 2 y 2 — 1,
you can easily show that
c 2 (p' x 2 +Py +Pf) - E 12 = c 2 (p 2 + p 2 +p 2 ,) - E 2 . (G.18)
Evidently this quantity c 2 p 2 — E 2 is not changed by a Lorentz transfor¬
mation. It is often called the invariant four-momentum (even though it
has dimensions of energy squared). It has the same value in every frame
of reference, including the particle’s rest frame. In the rest frame the par¬
ticle’s momentum is zero and its energy E is just moc 2 . The invariant
four-momentum is therefore —m 2 f A . It follows that in any other frame
E 2 = c 2 p 2 + m 2 0 c A . (G.19)
The invariant constructed in the same way with Eq. (G.2) is
(x B — x A y + (y B ~ Ja)” + ( z B — ZaY — c 2 (t B — t A y. (G.20)
Two events, A and B, for which this quantity is positive are said to have a
spacelike separation. It is always possible to find a frame in which they
are simultaneous. If the invariant is negative, the events have a timelike
separation. In that case a frame exists in which they occur at different
times, but at the same place. If this “invariant interval” is zero, the two
events can be connected by a flash of light.
Radiation by an
accelerated charge
A particle with charge q has been moving in a straight line at constant
speed i'o for a long time. It runs into something, let us imagine, and
in a short period of constant deceleration, of duration r, the particle is
brought to rest. The graph of velocity versus time in Fig. H.l describes
its motion. What must the electric field of this particle look like after
that? Figure H.2 shows how to derive it.
We shall assume that vo is small compared with c. Let t = 0 be
the instant the deceleration began, and let x = 0 be the position of the
particle at that instant. By the time the particle has completely stopped
it will have moved a little farther on, to x = vqx/2. That distance, indi¬
cated in Fig. H.2, is small compared with the other distances that will be
involved.
We now examine the electric field at a time t — T r. Observers
farther away from the origin than R = cT cannot have learned that the
particle was decelerated. Throughout that region, region I in Fig. H.2, the
field must be that of a charge that has been moving and is still moving at
the constant speed vq. That field, as we discovered in Section 5.7, appears
to emanate from the present position of the charge, which for an observer
anywhere in region I is the point x = vqT on the x axis. That is where the
particle would be now if it hadn’t been decelerated. On the other hand,
for any observer whose distance from the origin is less than c(T — r),
that is to say, for any observer in region II, the field is that of a charge at
rest close to the origin (actually at x = vqx/2).
What must the field be like in the transition region, the spherical
shell of thickness cr? Gauss’s law provides the key. A field line such
as AB lies on a cone around the x axis that includes a certain amount
of flux from the charge q. If CD makes the same angle 0 with the axis.
Radiation by an accelerated charge
813
the cone on which it lies includes that same amount of flux. (Because
i'o is small, the relativistic compression of field lines visible in Fig. 5.15
and Fig. 5.19 is here negligible.) Hence AB and CD must be parts of the
same field line, connected by a segment BC. This tells us the direction
of the field E within the shell; it is the direction of the line segment
BC. This field E within the shell has both a radial component E r and a
transverse component Eg. From the geometry of the figure their ratio is
easily found:
Eg vqT sin0
E r CT
(H.l)
V
Vo
\i_1
t = 0 t= T t=T
Figure H.l.
Velocity-time diagram for a particle that traveled
at constant speed v 0 until t = 0. It then
experienced a constant negative acceleration of
magnitude a = v 0 /t, which brought it to rest at
time t = r. We assume v 0 is small compared
with c.
Now E r must have the same value within the shell thickness that it does
in region II near B. (Gauss’s law again!) Therefore E r = q/AneqR 2 =
q/Aneqc 2 T 2 , and substituting this into Eq. (H.l) we obtain
Eg
vqT sin ft
-- E r
C X
qv o sin 9
Aneoc 3 Tr
(H.2)
Figure H.2.
Space diagram for the instant t = T » r, a long
time after the particle has stopped. For
observers in region I, the field must be that of a
charge located at the position x = v 0 T; for
observers in region II, it is that of a particle at
rest close to the origin. The transition region is a
shell of thickness ex.
814
Radiation by an accelerated charge
But v'o/r = a, the magnitude of the (negative) acceleration, and cT = R,
so our result can be written as follows:
qa sin 0
6 4jreoc 2 R
(H.3)
A remarkable fact is here revealed: Eg is proportional to 1 /R, not to
1 /R 2 ! As time goes on and R increases, the transverse field Eg will even¬
tually become very much stronger than E r . Accompanying this transverse
(that is, perpendicular to R) electric field will be a magnetic field of
strength Eg/c perpendicular to both R and E. This is a general property
of an electromagnetic wave, explained in Chapter 9.
Let us calculate the energy stored in the transverse electric field
above, in the whole spherical shell. The energy density is
eo£f _ g 2 a 2 sin 2 6
2 32tt 2 € 0 R 2 c 4 '
The volume of the shell is AitRrcx, and the average value of sin 2 0 over a
sphere 1 is 2/3. The total energy of the transverse electric field is therefore
2 2 q 2 a 2 q 2 a 2 x
-AttR-cx -^^—r = —- T .
3 32n 2 eoR 2 c 4 l2ixeoc 3
(H.5)
To this we must add an equal amount (see Section 9.6.1) for the energy
stored in the transverse magnetic field:
2 2
q ax
Total energy in transverse electromagnetic field = --. (H.6)
67T6oC j
The radius R has canceled out. This amount of energy simply travels
outward, undiminished, with speed c from the site of the deceleration.
Since r is the duration of the deceleration, and is also the duration of the
electromagnetic pulse a distant observer measures, we can say that the
power radiated during the acceleration process was
Brad —
2 1
q a
6jreo c 3
(H.7)
As it is the square of the instantaneous acceleration that appears in
Eq. (H.7), it doesn’t matter whether a is positive or negative. Of course
it ought not to, for stopping in one inertial frame could be starting in
1 Our polar axis in Fig. H.2 is the x axis: cos 2 8 = x 2 /R 2 . With a bar denoting an
average over the sphere, x 2 = y 2 = z 2 = R 2 j 3. Hence cos 2 8 = 1 /3, and
sin 2 8=1 — cos 2 8 = 2/3. Or you can just do an integral; the area of a circular strip
around the x axis is proportional to sin 8, so you end up integrating sin 3 8.
Exercises
815
another. Speaking of different frames, P rdl i itself turns out to be Lorentz-
invariant, which is sometimes very handy. That is because P rat j is energy /
time, and energy transforms like time, each being the fourth component
of a four-vector, as noted in Appendix G.
We have here a more general result than we might have expected.
Equation (H.7) correctly gives the instantaneous rate of radiation of
energy by a charged particle moving with variable acceleration - for
instance, a particle vibrating in simple harmonic motion. It applies to a
wide variety of radiating systems from radio antennas to atoms and nuclei.
Exercises
H.l Ratio of energies *
An electron moving initially at constant (nonrelativistic) speed v
is brought to rest with uniform deceleration a lasting for a time
t — v/a. Compare the electromagnetic energy radiated during the
deceleration with the electron’s initial kinetic energy. Express the
ratio in terms of two lengths, the distance light travels in time t and
the classical electron radius ro, defined as e 2 /4 jt eomc 2 .
H.2 Simple harmonic moton **
An elastically bound electron vibrates in simple harmonic motion
at frequency o> with amplitude A.
(a) Find the average rate of loss of energy by radiation.
(b) If no energy is supplied to make up the loss, how long will it
take for the oscillator’s energy to fall to 1 /e of its initial value?
(Answer: 6 jt eomc 3 / e 2 or .)
H.3 Thompson scattering **
A plane electromagnetic wave with frequency w and electric field
amplitude Eq is incident on an isolated electron. In the resulting
sinusoidal oscillation of the electron the maximum acceleration
is Eoe/m (the maximum force divided by m). How much power
is radiated by this oscillating charge, averaged over many cycles?
(Note that it is independent of the frequency co.) Divide this aver¬
age radiated power by eoE^c/2, the average power density (power
per unit area of wavefront) in the incident wave. This gives a con¬
stant a with the dimensions of area, called a scattering cross sec¬
tion. The energy radiated, or scattered, by the electron, and thus
lost from the plane wave, is equivalent to that falling on an area er.
(The case here considered, involving a free electron moving non-
relativistically, is often called Thomson scattering after J. J. Thom¬
son, the discoverer of the electron, who first calculated it.)
H.4 Synchrotron radiation **
Our master formula, Eq. (H.7), is useful for relativistically mov¬
ing particles, even though we assumed vq <$C c in the derivation.
816
Radiation by an accelerated charge
All we have to do is transform to an inertial frame F' in which
the particle in question is, at least temporarily, moving slowly,
apply Eq. (H.7) in that frame, then transform back to any frame
we choose. Consider a highly relativistic electron (y 1) moving
perpendicular to a magnetic field B. It is continually accelerated
perpendicular to the field, and must radiate. At what rate does
it lose energy? To answer this, transform to a frame F' moving
momentarily along with the electron, find E' in that frame, and / J ' ad .
Now show that, because power is (energy)/(time), P mi j = P' ad . This
radiation is generally called synchrotron radiation. (Answer: P rdc \ =
y 2 e 4 B 2 /6jt€om 2 c.)
The metal lead is a moderately good conductor at room temperature. Its
resistivity, like that of other pure metals, varies approximately in pro¬
portion to the absolute temperature. As a lead wire is cooled to 15 K its
resistance falls to about 1 /20 of its value at room temperature, and the
resistance continues to decrease as the temperature is lowered further.
But as the temperature 7.22 K is passed, there occurs without forewarn¬
ing a startling change: the electrical resistance of the lead wire van¬
ishes! So small does it become that a current flowing in a closed ring
of lead wire colder than 7.22 K - a current that would ordinarily die out
in much less than a microsecond - will flow for years without meas¬
urably decreasing. This phenomenon has been directly demonstrated.
Other experiments indicate that such a current could persist for billions
of years. One can hardly quibble with the flat statement that the resistiv¬
ity is zero. Evidently something quite different from ordinary electrical
conduction occurs in lead below 7.22 K. We call it superconductivity.
Superconductivity was discovered in 1911 by the great Dutch low-
temperature experimenter Kamerlingh Onnes. He observed it first in mer¬
cury, for which the critical temperature is 4.16 K. Since then hundreds
of elements, alloys, and compounds have been found to become super¬
conductors. Their individual critical temperatures range from roughly a
millikelvin up to the highest yet discovered, 138 K. Curiously, among
the elements that do not become superconducting are some of the best
normal conductors such as silver, copper, and the alkali metals.
Superconductivity is essentially a quantum-mechanical phenomenon,
and a rather subtle one at that. The freely flowing electric current consists
of electrons in perfectly orderly motion. Like the motion of an electron
in an atom, this electron flow is immune to small disturbances - and for
Superconductivity
818
Superconductivity
a similar reason: a finite amount of energy would be required to make
any change in the state of motion. It is something like the situation in
an insulator in which all the levels in the valence band are occupied and
separated by an energy gap from the higher energy levels in the conduction
band. But unlike electrons filling the valence band, which must in total
give exactly zero net flow, the lowest energy state of the superconducting
electrons can have a net electron velocity, hence current flow, in some
direction. Why should such a strange state become possible below a
certain critical temperature? We can’t explain that here. 1 It involves the
interaction of the conduction electrons not only with each other, but also
with the whole lattice of positive ions through which they are moving.
That is why different substances can have different critical temperatures,
and why some substances are expected to remain normal conductors right
down to absolute zero.
In the physics of superconductivity, magnetic fields are even more
important than you might expect. We must state at once that the phenom¬
ena of superconductivity in no way violate Maxwell’s equations. Thus
the persistent current that can flow in a ring of superconducting wire is
a direct consequence of Faraday’s law of induction, given that the resis¬
tance of the ring is really zero. For if we start with a certain amount of
flux <t>o threading the ring, then because / E ■ ds around the ring remains
always zero (otherwise there would be infinite current due to the zero
resistance), di>/dt must be zero. The flux cannot change; the current I in
the ring will automatically assume whatever value is necessary to main¬
tain the flux at <t>o- Figure 1.1 outlines a simple demonstration of this, and
shows how a persistent current can be established in an isolated super¬
conducting circuit.
Superconductors can be divided into two types. In Type 1 supercon¬
ductors, the magnetic field inside the material itself (except very near the
surface) is always zero. That is not a consequence of Maxwell’s equa¬
tions, but a property of the superconducting state, as fundamental, and
once as baffling, a puzzle as the absence of resistance. The condition
B = 0 inside the bulk of a Type 1 superconductor is automatically main¬
tained by currents flowing in a thin surface layer. In Type 2 supercon¬
ductors, quantized magnetic flux tubes may exist for a certain range of
temperature and external magnetic field. These tubes are surrounded by
vortices of current (essentially little solenoids) which allow the magnetic
field to be zero in the rest of the material. Outside the flux tubes the
material is superconducting.
A strong magnetic field destroys superconductivity, although Type 2
superconductors generally can tolerate much larger magnetic fields than
1 The abrupt emergence of a state of order at a certain critical temperature reminds us
of the spontaneous alignment of electron spins that occurs in iron below its Curie
temperature (mentioned in Section 11.11). Such cooperative phenomena always
involve a large number of mutually interacting particles. A more familiar cooperative
phenomenon is the freezing of water, also characterized by a well-defined critical
temperature.
Superconductivity
819
Ring of solder (lead-tin
alloy); normal conductor;
current zero; permanent
magnet causes flux O 0
through ring.
(c)
r\
/-\
s~\
Ring cooled below its critical
temperature. (Some helium
has boiled away.) Flux through
ring unchanged. Ring is now
a superconductor.
Magnet removed. Persistent
current I now flows in ring
to maintain flux at value <D 0 .
Compass needle responds to
field of persistent current.
Figure 1.1.
Establishing a persistent current in a
superconducting ring. The ring is made of
ordinary solder, a lead-tin alloy, (a) The ring, not
yet cooled, is a normal conductor with ohmic
resistance. Bringing up the permanent magnet
will induce a current in the ring that will quickly
die out, leaving the magnetic flux from the
magnet, in amount <i>, passing through the ring,
(b) The helium bath is raised without altering the
relative position of the ring and the permanent
magnet. The ring, now cooled below its critical
temperature, is a superconductor with
resistance zero, (c) The magnet is removed.
The flux through the zero resistance ring cannot
change. It is maintained at the value <f> by a
current in the ring that will flow as long as the
ring remains below the critical temperature. The
magnetic field of the persistent current can be
demonstrated with the compass.
Type 1. None of the superconductors known before 1957 could stand
more than a few hundred gauss. That discouraged practical applications
of zero-resistance conductors. One could not pass a large current through
a superconducting wire because the magnetic field of the current itself
would destroy the superconducting state. But then a number of Type 2
superconductors were discovered that could preserve zero resistance in
fields up to 10 tesla or more. A widely used Type 2 superconductor is
820
Superconductivity
an alloy of niobium and tin that has a critical temperature of 18 K and
if cooled to 4 K remains superconducting in fields up to 25 tesla. Type
2 superconducting solenoids are now common that produce steady mag¬
netic fields of 20 tesla without any cost in power other than that incident
to their refrigeration. Uses of superconductors include magnetic reso¬
nance imaging (MRI) machines (which are based on the physics dis¬
cussed in Appendix J) and particle accelerators. There are also good
prospects for the widespread use of superconductors in large electrical
machinery, maglev trains, and the long-distance transmission of electri¬
cal energy.
In addition to the critical magnetic field, the critical temperature is
also a factor in determining the large-scale utility of a superconductor. In
particular, a critical temperature higher than 77 K allows relatively cheap
cooling with liquid nitrogen (as opposed to liquid helium at 4 K). Prior to
1986, the highest known critical temperature was 23 K. Then a new type
of superconductor (a copper oxide, or cuprate ) was observed with a criti¬
cal temperature of 30 K. The record critical temperature was soon pushed
to 138 K. These superconductors are called high-temperature supercon¬
ductors. Unfortunately, although they are cheaper to cool, their utility is
limited because they tend to be brittle and hence difficult to shape into
wires. However, in 2008 a new family of high-temperature superconduc¬
tors was discovered, with iron as a common element. This family is more
ductile than cuprates, but the highest known critical temperature is 55 K.
The hope is that this will eventually cross the 77 K threshold.
The mechanism that leads to high-temperature superconductivity
is more complex than the mechanism for low-temperature supercon¬
ductivity. In contrast with the well-established BCS theory (named after
Bardeen, Cooper, and Schrieffer; formulated in 1957) for low-temperature
superconductors, a complete theory of high-temperature superconduc¬
tors does not yet exist. All known high-temperature superconductors are
Type 2, but not all Type 2 superconductors are high-temperature. Indeed,
/ow-temperature Type 2 superconductors (being both ductile and tolerant
of large magnetic fields) are the ones presently used in MRI machines
and other large-scale applications.
At the other end of the scale, the quantum physics of superconduc¬
tivity makes possible electrical measurements of unprecedented sensitiv¬
ity and accuracy - including the standardization of the volt in terms of an
easily measured oscillation frequency. To the physicist, superconductiv¬
ity is a fascinating large-scale manifestation of quantum mechanics. We
can trace the permanent magnetism of the magnet in Fig. 1.1 down to the
intrinsic magnetic moment of a spinning electron - a kind of supercur¬
rent in a circuit less than 10“ 10 m in size. The ring of solder wire with
the persistent current flowing in it is, in some sense, like a gigantic atom,
the motion of its associated electrons, numerous as they are, marshaled
into the perfectly ordered behavior of a single quantum state.
The electron has angular momentum of spin, J. Its magnitude is always
the same, hi An, or 5.273- 10“ 35 kgm 2 /s. Associated with the axis of
spin is a magnetic dipole moment p of magnitude 0.9285 ■ 10 -23 joule/
tesla (see Section 11.6). An electron in a magnetic field experiences
a torque tending to align the magnetic dipole in the field direction. It
responds like any rapidly spinning gyroscope: instead of lining up with
the field, the spin axis precesses around the field direction. Let us see
why any spinning magnet does this. In Fig. J.l the magnetic moment p
is shown pointing opposite to the angular momentum J, as it would for
a negatively charged body like an electron. The magnetic field B (the
field of some solenoid or magnet not shown) causes a torque equal to
fi x B. This torque is a vector in the negative x direction at the time
of our picture. Its magnitude is given by Eq. (11.48); it is pB sin 0. In
a short time At, the torque adds to the angular momentum of our top a
vector increment AJ in the direction of the torque vector and of magni¬
tude pB sin 0 At. The horizontal component of J, in magnitude J sin 0,
is thereby rotated through a small angle A \j/ given by
Magnetic resonance
A l/r
AJ
J sin 9
pB At
J
(J.l)
As this continues, the upper end of the vector J will simply move around
the circle with constant angular velocity &> p :
co
P —
Ai fr
At
pB
(J-2)
822
Magnetic resonance
Figure J.l.
The precession of a magnetic top in an external
field. The angular momentum of spin J and the
magnetic dipole moment /t are oppositely
directed, as they would be for a negatively
charged rotor.
This is the rate of precession of the axis of spin. Note that it is the same
for any angle of tip; sin 0 has canceled out.
For the electron, /r// has the value 1.761 • 10 11 s _1 tesla _1 . In a field
of 1 gauss (10~ 4 tesla) the spin vector precesses at 1.761 • 10 7 radians/s,
or 2.80- 10 6 revolutions per second. The proton has exactly the same
intrinsic spin angular momentum as the electron, h/An , but the associ¬
ated magnetic moment is smaller. That is to be expected since the mass
of the proton is 1836 times the mass of the electron; as in the case of
orbital angular momentum (see Eq. (11.29)), the magnetic moment of an
elementary particle with spin ought to be inversely proportional to its
mass, other things being equal. Actually the proton’s magnetic moment
is 1.411 ■ 10~ 26 joule/tesla, only about 660 times smaller than the electron
moment, which shows that the proton is in some way a composite par¬
ticle. In a field of 1 gauss the proton spin precesses at 4258 revolutions
per second. About 40 percent of the stable atomic nuclei have intrinsic
angular momenta and associated magnetic dipole moments.
We can detect the precession of magnetic dipole moments through
their influence on an electric circuit. Imagine a proton in a magnetic field
B, with its spin axis perpendicular to the field, and surrounded by a small
coil of wire, as in Fig. J.2. The precession of the proton causes some
alternating flux through the coil, as would the end-over-end rotation of a
little bar magnet. A voltage alternating at the precession frequency will
be induced in the coil. As you might expect, the voltage thus induced
by a single proton would be much too feeble to detect. But it is easy to
provide more protons - 1 cm 3 of water contains about 7 ■ 10 22 protons
(we’re concerned with the two hydrogen atoms in each water molecule),
and all of them will precess at the same frequency. Unfortunately they
will not all be pointing in the same direction at the same instant. In fact,
their spin axes and magnetic moments will be distributed so uniformly
over all possible directions that their fields will very nearly cancel one
another. But not quite, if we introduce another step. If we apply a strong
magnetic field B to water, for several seconds there will develop a slight
excess of proton moments pointing in the direction of B, the direction
they energetically favor. The fractional excess will be fiB/kT in order of
magnitude, as in ordinary paramagnetism. It may be no more than one
in a million, but these uncanceled moments, if they are now caused to
precess in our coil, will induce an observable signal.
A simple method for observing nuclear spin precession in weak
fields, such as the earth’s field, is described in Fig. J.3. Many other
schemes are used to observe the spin precession of electrons and of
Figure J.2.
A processing magnetic dipole moment at the center of a coil causes a
periodic change in the flux through the coil, inducing an alternating
electromotive force in the coil. Note that the flux from the dipole m that
links the coil is that which loops around outside it. See Exercise J.l.
Magnetic resonance
823
nuclei. They generally involve a combination of a steady magnetic field
and oscillating magnetic fields with frequency in the neighborhood of
co p . For electron spins (electron paramagnetic resonance, or EPR) the
frequencies are typically several thousand megahertz, while for nuclear
spins (nuclear magnetic resonance, or NMR) they are several tens of
megahertz. The exact frequency of precession, or resonance, in a given
applied field can be slightly shifted by magnetic interactions within a
molecule. This has made NMR, in particular, useful in chemistry. The
position of a proton in a complex molecule can often be deduced from
the small shift in its precession frequency.
Magnetic fields easily penetrate ordinary nonmagnetic materials,
and that includes alternating magnetic fields if their frequency or the
electric conductivity of the material is not too great. A steady field of
2000 gauss applied to the bottle of water in our example would cause
any proton polarization to precess at a frequency of 8.516-10 6 revolu¬
tions per second. The field of the precessing moments would induce a
signal of 8.516 MHz frequency in the coil outside the bottle. This applies
as well to the human body, which, viewed as a dielectric, is simply an
assembly of more or less watery objects. In NMR imaging (or magnetic
resonance imaging, MRI) the interior of the body is mapped by means of
nuclear magnetic resonance. The concentration of hydrogen atoms at a
Figure J.3.
Apparatus for observing proton spin precession
in the earth's field s e . A bottle of water is
surrounded by two orthogonal coils. With switch
S 2 open and switch Si closed, the large solenoid
creates a strong magnetic field s 0 . As in
ordinary paramagnetism (Section 11.6), the
energy is lowered if the dipoles point in the
direction of the field, but thermal agitation
causes disorder. Our dipoles here are the
protons (hydrogen nuclei) in the molecules of
water. When thermal equilibrium has been
attained, which in this case takes several
seconds, the magnetization is what you would
get by lining up with the magnetic field the small
fraction /j.B 0 /kT of all the proton moments. We
now switch off the strong field s 0 and close
switch S 2 to connect the coil around the bottle to
the amplifier. The magnetic moment m now
precesses in the xy plane around the remaining,
relatively weak, magnetic field s e , with
precession frequency given by Eq. (J.2). The
alternating y component of the rotating vector m
induces an alternating voltage in the coil which
can be amplified and observed. From its
frequency, B e can be very precisely determined.
This signal itself will die away in a few seconds
as thermal agitation destroys the magnetization
the strong field So had brought about. Magnetic
resonance magnetometers of this and other
types are used by geophysicists to explore the
earth’s field, and even by archaeologists to
locate buried artifacts.
824
Magnetic resonance
particular location is revealed by the radiofrequency signal induced in an
external coil by the precessing protons. The location of the source within
the body can be inferred from the precise frequency of the signal if the
steady field B, which determines the frequency according to Eq. (J.2),
varies spatially with a known gradient.
Exercises
J.l Emffrom a proton **
At the center of the four-turn coil of radius a in Fig. J.2 is a sin¬
gle proton, precessing at angular rate a> p . Derive a formula for
the amplitude of the induced alternating electromotive force in the
coil, given that the proton moment is 1.411 • 10 -26 joule/tesla.
J.2 Emffrom a bottle ***
(a) If the bottle in Fig. J.3 contains 200 cm 3 of H 2 O at room tem¬
perature, and if the field Bq is 1000 gauss, how large is the net
magnetic moment m?
(b) Using the result of Exercise J.l, make a rough estimate of the
signal voltage available from a coil of 500 turns and 4 cm
radius when the field strength B e is 0.4 gauss.
K.1 Fundamental constants
speed of light
c
2.998 ■ 10 8 m/s
elementary charge
e
1.602-10“ 19 C
4.803- 10“ 10 esu
electron mass
m e
9.109 • 10 -31 kg
proton mass
m p
1.673 • 10- 27 kg
Avogadro’s number
Na
6.022- 10“ 23 mole -1
Boltzmann constant
k
1.381 • 10 -23 J/K
Planck constant
h
6.626 ■ 10 -34 J s
gravitational constant
G
6.674-10 -11 m 3 /(kg s 2 )
electron magnetic moment
Me
9.285-10 -24 J/T
proton magnetic moment
Mp
1.411 • 10 -26 J/T
permittivity of free space
eo
8.854- 10 -12 C 2 /(Nm 2 )
permeability of free space
Mo
1.257-10 -6 Tm/A
The exact numerical value of /tq is 4 tt ■ 10 7 (by definition).
Helpful
formulas/facts
The exact numerical value of eo is (4 jt ■ [3] 2 ■ 10 9 ) , where [3] =
2.99792458 (see Appendix E).
826
Helpful formulas/facts
K.2 Integral table
/ dx 1 /x\
— - T = - tan ( - )
x-+ r- r \r /
J
f
f
f
f
xr + r 2
dx
Vl — X 2
dx
Vx 2 — 1
dx
six 2 + a 2
dx
( a 2 + x 2 ) 3 / 2
I' axdx
x" In ^ ^ dx ■
/«-*:
/a-*
J sin 3 xdx
sin x
: In (x + s/x 2 — 1)
: In (^/x 2 + a 2 + x^
X
a 2 (a 2 + x 2 ) 1 / 2
: x In x — x
yH-\- 1
(n + l) 2
— (x+ l)e~
x / a\
-In ( - )
n + 1 Vx/
— —(x -j - 2x 4“ 2)e
cos 3 x
= — cosx-
/
COS 3 xdx =
sin 3 x
/
/
dx
cosx
dx
sinx
In I 1 + sinjc )
= In
/
/
cos x dx
cosx
1 — cosx
sinx
sinx
/
/
(1 — a 2 cos 2 x) 3 / 2
sin x dx
(1 — a 2 sin 2 x) 3 / 2
cos x dx
(1 — b 2 sin 2 (x — a)) 3 '"
sin x(o cos x — b)dx
(a 2 4- b 2 - 2a£>cosx) 3 / 2 Z?Va 2 4- fe 2 - 2a/?cosx
(1 — a 2 )Vl — a 2 cos 2 x
— cosx
(1 — a 2 )\/1 — a 2 sin 2 x
(2 — b 2 ) sinx 4- d 2 sin(2a — x)
2(1 — b 2 )\J 1 — b 2 sin 2 (a — x)
—a 4- b cosx
(K.1)
(K.2)
(K.3)
(K.4)
(K.5)
(K.6)
(K.7)
(K.8)
(K.9)
(K.10)
(K.ll)
(K.12)
(K.13)
(K.14)
(K.15)
(K.16)
(K.17)
K.4 Taylor series
827
K.3 Vector identities
V ■ (V x A) = 0
V ■ (fA) =/V ■ A + A • V/
V • (A x B) = B ■ (V x A) - A • (V x B)
V x (V/) = 0
V x (fA ) =/V x A + (V/) x A
V x (V x A) = V(V ■ A) - V 2 A
V x (A x B) = A(V • B) - B(V • A) + (B • V)A - (A ■ V)B
A x (B x C) = B(A • C) - C(A • B)
V(A ■ B) = (A • V)B + (B ■ V)A + A x (V x B) + B x (V x A)
K.4 Taylor series
The general form of a Taylor series is
/ f"{x o) 9 f'"{x o) 9
fix 0 + x) = /(x 0 ) +/'(xo)x + 2, X 2 + ^ x 3 + • • • . (K.18)
This equality can be verified by taking successive derivatives and then
setting x = 0. For example, taking the first derivative and then setting
x = 0 gives f (xf) on the left, and also/'(xo) on the right, because the
first term is a constant and gives zero when differentiated, the second
term gives/'(xo), and all the rest of the terms give zero once we set x— 0
because they all contain at least one power of x. Likewise, if we take
the second derivative of each side and then set x = 0, we obtain f"( xq)
on both sides. And so on for all derivatives. Therefore, since the two
functions on each side of Eq. (K.18) are equal at x = 0 and also have
their nth derivatives equal at x = 0 for all n, they must in fact be the
same function (assuming that they are nicely behaved functions, which
we generally assume in physics).
Some specific Taylor series that come up often are listed below;
they are all expanded around xo = 0. We use these series countless times
throughout this book when checking how expressions behave in the
limit of some small quantity. The series are all derivable via Eq. (K.18),
but sometimes there are quicker ways of obtaining them. For example,
Eq. (K.20) is most easily obtained by taking the derivative of Eq. (K.19),
which itself is simply the sum of a geometric series.
—= 1 +X + X 2 +x 3 + ••• (K.19)
1 —x
„ 1 = 1 + 2x + 3x 2 + 4x 3 + • • ■ (K.20)
(1 - x)-
2 3
X X
ln(l -x) = -x- -- — -
(K.21)
828
Helpful formulas/facts
x X
e = 1 + x H-1-1-
2! 3!
(K.22)
x 2 x 4
cosx =1 -1-
2! 4!
(K.23)
Jt 3 x 3
sinx = x -1-•••
3! 5!
(K.24)
2
Vl+X=l + ^-y+-"
(K.25)
1 i x 3x 2
2 + 8 +
(K.26)
(1 + x)" = 1 +nx+ + ("3 V 3 ^ -
(K.27)
K.5 Complex numbers
The imaginary number i is defined to be the number for which i 2 =
— 1. (Of course, —i also has its square equal to —1.) A general complex
number z with both real and imaginary parts can be written in the form
a + bi, where a and b are real numbers. Such a number can be described
by the point (a, b ) in the complex plane, with the x and y axes being the
real and imaginary axes, respectively.
The most important formula involving complex numbers is
e ,e = cos 0 + i sin 0. (K.28)
This can quickly be proved by writing out the Taylor series for both sides.
Using Eq. (K.22), the first, third, fifth, etc. terms on the left-hand side of
Eq. (K.28) are real, and from Eq. (K.23) their sum is cos 0. Similarly, the
second, fourth, sixth, etc. terms are imaginary, and from Eq. (K.24) their
sum is i sin 0. Writing it all out, we have
e ie = 1
• id
{ley aoy {wy
2 !
, e- e 4
= i — + —
2! 4!
3!
4!
aey
■+
5!
e 3 e 5
+ l[d ~ 3! + 5!
cos (
(K.29)
as desired.
Letting 6 —»■ — 6 in Eq. (K.28) yields e~ ,e = cos 6 — isind. Com¬
bining this with Eq. (K.28) allows us to solve for cos 0 and sin 6 in terms
of the complex exponentials:
e ie + e -w
cos 9 =
2
sind =
2 i
(KAO)
K.6 Trigonometric identities
829
A complex number z described by the Cartesian coordinates (a, b) in
the complex plane can also be described by the polar coordinates ( r , 9).
The radius r and angle 9 are given by the usual relation between Carte¬
sian and polar coordinates (see Fig. K.l),
r = V a 2 + b 2 and 9 = tan -1 (b/a). (K.31)
Using Eq. (K.28), we can write z in polar form as
a + bi = (rcos9) + (rsm9)i = r(cos9 + isin@) = re' 0 . (K.32)
y
We see that the quantity in the exponent (excluding the i) equals the angle Figure K.l.
of the vector in the complex plane. Cartesian and polar coordinates in the complex
The complex conjugate of z, denoted by z* (or by z), is defined to P^ ane -
be z* = a — bi, or equivalently z* = re - ' 0 . It is obtained by reflecting
the Cartesian point (a, b ) across the real axis. Note that either of these
expressions for z* implies that r can be written as r = sfzz*. The radius
r is known as the magnitude or absolute value of z, and is commonly
denoted by |z|. The complex conjugate of a product is the product of the
complex conjugates, that is, (zizf)* = zjzJj- You can quickly verify this
by writing z t and z .2 in polar form. The Cartesian form works too, but that
takes a little longer. The same result holds for the quotient of two com¬
plex
numbers.
As an example of the use of Eq. (K.28), we can quickly derive the
double-angle formulas for sine and cosine. We have
cos 29 + i&\t\29 = e' 20 = (e' 0 ) 2 = (cost? + isin0) 2
= (cos 2 9 — sin 2 9) + i( 2 sin 9 cos 9). (K.33)
Equating the real parts of the expressions on either end of this equation
gives cos 29 = cos 2 9 — sin 2 9. And equating the imaginary parts gives
sin 29 = 2 sin 9 cos 9. This method easily generalizes to other trig sum
formulas.
K.6 Trigonometric identities
sin 29 = 2 sin 9 cos 9, cos 29 — cos 2 9 — sin 2 (
sin(a + f) — sin a cos ft + cos a sin ft
cos(a + f) = cos a cos f — sin a sin f
tan a + tan B
tan(a + ft) =
1 — tan a tan i
(K.34)
(K.35)
(K.36)
(K.37)
830
Helpful formulas/facts
cos — = ±
2
1 + cos 6
6 /1 — cos(
sin — — ±. -
2 V 2
/1 — cos 6
tan — = ±,,
2 V 1 + cos 9
1 — cos 6
sin0
1 + cos 6
(K.38)
(K.39)
The hyperbolic trig functions are defined by analogy with Eq. (K.30),
with the fs omitted:
e x + e~ x
coshx = -,
2
cosh 2 x —
— cosh x — sinh x,
dx
sinhx = -
2
sinh 2 x = 1
d
— sinh x = cosh x
dx
(K.40)
(K.41)
(K.42)
Andrews, M. (1997). Equilibrium charge density on a conducting needle. Am. J.
Phys., 65, 846-850.
Assis, A. K. T., Rodrigues, W. A., Jr., and Mania, A. J. (1999). The electric field
outside a stationary resistive wire carrying a constant current. Found. Phys.,
29, 729-753.
Auty, R. R and Cole, R. H. (1952). Dielectric properties of ice and solid D 2 O.
J. Chem. Phys., 20, 1309-1314.
Blakemore, R. P. and Frankel, R. B. (1981). Magnetic navigation in bacteria. Sci.
Am., 245, (6), 58-65.
Bloomfield, L.A. (2010). How Things Work, 4th edn. (New York: John Wiley &
Sons).
Boos, F. L., Jr. (1984). More on the Feynman’s disk paradox. Am. J. Phys., 52,
756-757.
Bose, S. K. and Scott, G. K. (1985). On the magnetic field of current through a
hollow cylinder. Am. J. Phys., 53, 584-586.
Crandall, R. E. (1983). Photon mass experiment. Am. J. Phys., 51, 698-702.
Crawford, F. S. (1992). Mutual inductance Myi = M 21 : an elementary deriva¬
tion. Am. J. Phys., 60. 186.
Crosignani, B. and Di Porto, P. (1977). Energy of a charge system in an equilib¬
rium configuration. Am. J. Phys., 45, 876.
Davis, L., Jr., Goldhaber, A. S., and Nieto, M. M. (1975). Limit on the
photon mass deduced from Pioneer-10 observations of Jupiter’s magnetic
field. Phys. Rev. Lett., 35, 1402-1405.
Faraday, M. (1839). Experimental Researches in Electricity (London: R. and
J. E. Taylor).
Feynman, R. P, Leighton, R. B., and Sands, M. (1977). The Feynman Lectures
on Physics, vol. II (Reading, MA: Addision-Wesley).
Friedberg, R. (1993). The electrostatics and magnetostatics of a conducting disk.
Am. J. Phys., 61, 1084-1096.
References
832
References
Galili, I. and Goihbarg, E. (2005). Energy transfer in electrical circuits: a quali¬
tative account. Am. J. Phys., 73, 141-144.
Goldhaber, A. S. and Nieto, M. M. (1971). Terrestrial and extraterrestrial limits
on the photon mass. Rev. Mod. Phys., 43, 277-296.
Good, R. H. (1997). Comment on “Charge density on a conducting needle,” by
David J. Griffiths and Ye Li [Am. J. Phys. 64(6), 706-714 (1996)]. Am. J.
Phys., 65, 155-156.
Griffiths, D. J. and Heald, M. A. (1991). Time-dependent generalizations of the
Biot-Savart and Coulomb laws. Am. J. Phys., 59, 111-117.
Griffiths, D. J. and Li, Y. (1996). Charge density on a conducting needle. Am. J.
Phys., 64. 706-714.
Hughes, V. W. (1964). In Chieu, H. Y. and Hoffman, W. F. (eds.), Gravitation
and Relativity. (New York: W. A. Benjamin), chap. 13.
Jackson, J. D. (1996). Surface charges on circuit wires and resistors play three
roles. Am. J. Phys., 64, 855-870.
Jefimenko, O. (1962). Demonstration of the electric fields of current-carrying
conductors. Am. J. Phys., 30, 19-21.
King, J. G. (1960). Search for a small charge carried by molecules. Phys. Rev.
Lett., 5, 562-565.
Macaulay, D. (1998). The New Way Things Work (Boston: Houghton Mifflin).
Marcus, A. (1941). The electric field associated with a steady current in long
cylindrical conductor. Am. J. Phys., 9, 225-226.
Maxwell, J. C. (1891). Treatise on Electricity and Magnetism, vol. I, 3rd
edn. (Oxford: Oxford University Press), chap. VII. (Reprinted New York:
Dover, 1954.)
Mermin, N. D. (1984a). Relativity without light. Am. J. Phys., 52, 119-124.
Mermin, N. D. (1984b). Letter to the editor. Am. J. Phys., 52, 967.
Nan-Xian, C. (1981). Symmetry between inside and outside effects of an electro¬
static shielding. Am. J. Phys., 49, 280-281.
O’Dell, S. L. and Zia, R. K. P. (1986). Classical and semiclassical diamagnetism:
a critique of treatment in elementary texts. Am. J. Phys., 54, 32-35.
Page, L. (1912). A derivation of the fundamental relations of electrodynamics
from those of electrostatics. Am. J. Sci., 34, 57-68.
Press, F. and Siever, R. (1978). Earth, 2nd edn. (New York: W. H. Freeman).
Priestly, J. (1767). The History and Present State of Electricity , vol. II, London.
Roberts, D. (1983). How batteries work: a gravitational analog. Am. J. Phys., 51,
829-831.
Romer, R. H. (1982). What do “voltmeters” measure? Faraday’s law in a multiply
connected region. Am. J. Phys., 50, 1089-1093.
Semon, M. D. and Taylor, J. R. (1996). Thoughts on the magnetic vector poten¬
tial. Am. J. Phys., 64, 1361-1369.
Smyth, C. P. (1955). Dielectric Behavior and Structure (New York: McGraw-
Hill).
Varney, R. N. and Fisher, L. H. (1980). Electromotive force: Volta’s forgotten
concept. Am. J. Phys., 48, 405-408.
Waage, H. M. (1964). Projection of electrostatic field lines. Am. J. Phys., 32, 388.
Whittaker, E. T. (1960). A History of the Theories of Aether and Electricity, vol. I
(New York: Harper), p. 266.
Williams, E. R., Faller, J. E., and Hill, H. A. (1971). New experimental test of
Coulomb’s law: a laboratory upper limit on the photon rest mass. Phys.
Rev. Lett., 26, 721-724.
Addition of velocities, relativistic, 808-809
additivity of interactions, 10, 13
admittance, 408-414
Alnico V, B-H curve for, 569
alternating current, 394-418
representation by complex number, 406-408
alternating-current circuit, 405-414
power and energy in, 415-418
alternating electromotive force, 395
alternator, 371
aluminum, doping of silicon with, 203-204
ammeter, 224
ammonia molecule, dipole moment of, 483
ampere (unit of current), 178, 283, 762—763, 790
Ampere, Andre-Marie, 2, 236, 238, 259, 531
Ampere’s law, 288
differential form, 291
amplitude modulation (AM), 455
Andrews, M., 640
angular momentum
conservation of, in changing magnetic field,
580
of electron spin, 546-547
Index
orbital, relation to magnetic moment, 541
precession of, 822-823
anode of vacuum diode, 181
antimatter, 3
antineutron, 3
antiproton, 3
Assis, A. K. T., 263
atom, electric current in, 540
atomic polarizability, 480-482
aurora borealis, 318
Auty, R. P., 505
B, magnetic field, 239, 278
and M, and H inside magnetized cylinder, 565
bacteria, magnetic, 571, 580
battery, lead-sulfuric acid, 209-212
B-H curve, 569-570
Biot-Savart law, 298, 435
Bitter plates, 320
Blakemore, R. R, 580
Bloomfield, L. A., 35
Bohr radius ciq, 55, 481, 544
Boltzmann factor, 202
Boltzmann’s constant k, 202, 503
Boos, F. L., 460
boost converter, 372
Bose, S. K., 305
bound and free charge, 497-498
arbitrariness of the distinction, 506-507
bound-charge current, 505-507
bound-charge density, 498
bound currents, 559-560
boundary of dielectric, change in E at, 494-495
boundary-value problem, 132, 151-153
bridge network, 208, 233
capacitance, 141—147
of cell membrane, 513
coefficients of, 148
of prolate spheroid, 171
units of, 142
illustrated, 145
capacitor, 141-147
dielectric-filled, 489-492
energy stored in, 149-151
parallel-plate, 143-144, 467
834
Index
capacitor (cont.)
uses of, 153
vacuum, 467
capacitor plate, force on, 151, 162
carbon monoxide molecule, dipole moment of,
483
cartesian coordinates, 791
cassette tape, 570
cathode of vacuum diode, 181
Cavendish, Henry, 11
centimeter (as unit of capacitance), 145
CH 3 OH (methanol) molecule, dipole moment
of, 483
charge
electric, see electric charge
magnetic, absence of, 529
in motion, see moving charge
charge density, linear, 28
charge distribution
cylindrical, field of, 83
electric, 20-22
moments of, 74, 471^-74
spherical, field of, 26-28
on a surface, 29
charged balloon, 32
charged disk, 68-71
field lines and equipotentials of, 72
potential of, 69
charged wire
potential of, 68
circuit breaker, 320
circuit element, 205
circuits
LR, 366-367
RC, 215-216
RLC, 389, 398, 410
alternating-current, 394-418
direct-current, 204-207
equivalent, 206
resonant, 388-394
circulation, 90
Clausius-Mossotti relation, 502
CO (carbon monoxide) molecule, dipole
moment of, 483
coefficients
of capacitance, 148
of potential, 148
coil
cylindrical (solenoid), magnetic field of,
300-303, 338
toroidal
energy stored in, 369
inductance of, 364
Cole, R. H., 505
comets, 454
compass needle, 239
complex exponential solutions, 402-405
complex-number representation of alternating
current, 406-408
complex numbers, review of, 828-829
conduction, electrical, 181—204
ionic, 189-195
in metals, 198-200
in semiconductors, 200-204
conduction band, 201-202
conductivity, electrical, 182-188
anisotropic, 182
of metals, 198-200
units for, 182
of various materials, 188, 195-197
conductors, electrical, 125-141
charged, system of, 128
properties of, 129
spherical, field around, 131
conformal mapping, 151
conservation of electric charge, 4—5,
180-181
distinguished from charge invariance, 242
conservative forces, 12
continuity equation, 181
copper, resistivity of, 188, 196-197
copper chloride, paramagnetism of, 526
corona discharge, 37
coulomb (SI unit of charge), 8 , 762
relation to esu, 9
Coulomb, Charles de, 10
Coulomb’s law, 7-11, 259
tests of, 10—11
Crandall, R. E., 11
Crawford, F. S., 378
critical damping, 394
Crosignani, B., 590
cross product (vector product) of two vectors,
238
Curie, Pierre, 566
Curie point, 566
curl, 90-99, 798-799
in Cartesian coordinates, 93-95, 100
physical meaning of, 95
curlmeter, 96
current density J, 177-180
current loop
magnetic dipole moment of, 534
magnetic field of, 531-535
torque on, 547
current ring, magnetic field of, 299
current sheet, 303-306
magnetic field of, 303-304
currents
alternating, 394-418
bound and free, 559-560
bound-charge, 505-507
displacement, 433-436
electric, see electric currents
fluctuations of, random, 195
curvilinear coordinates, 791-801
cylinder, magnetized, compared with cylinder
polarized, 557
cylindrical coordinates, 792
damped harmonic oscillator, 389
damped sinusoidal oscillation, 392
damping of resonant circuit, 388-394
critical, 394
Davis, L., Jr., 11
decay of proton, 6
decay time for earth’s magnetic field, 386
deer, flying, 102
“del” notation, 83, 95, 100
detergent, 510
deuterium molecule, 242
Di Porto, R, 590
diamagnetic substances, 526
diamagnetism, 527, 540, 546
of electron orbits, 545
diamond
crystal structure of, 200
wide band gap of, 203
dielectric constant k, 468
of various substances, 469
dielectric sphere in uniform field,
495-496
dielectrics, 467-471
Index
835
diode, 219
silicon junction, 229
vacuum, 181
dipole
comparison of electric and magnetic, 535-536
electric, see electric dipole
magnetic, see magnetic dipole
dipole moment
electric, see electric dipole moment
magnetic, see magnetic dipole moment
disk
conducting, field of, 140
charged, 68-72
displacement, electric, D, 499, 560-561
displacement current, 433-436
distribution of electric charge, 20-22
divergence, 78-79, 795-797
in Cartesian coordinates, 81-83, 100
divergence theorem, 79-80, 100
domains, magnetic, 567
doorbell, 321
doping of silicon, 203-204
dot product of two vectors, 12
dynamic random access memory (DRAM), 153
dynamo, 379, 386
dyne (Gaussian unit of force), 8
permittivity of free space, 8
Earnshaw’s theorem, 87
earth’s magnetic field, 280, 577
decay time of, 386
possible source of, 380
eddy-current braking, 370
Edison, Thomas, 419
Einstein, Albert, 2, 236, 281, 314
electret, 558
electric charge, 1—11, 242
additivity of, 10, 13
conservation of, 4—5, 180-181
distribution of, 20-22
free and bound, 497-498, 506-507
fundamental quantum of, 8
invariance of, 241-243
quantization of, 5—7, 242
sign of, 4
electric currents, 177—189
and charge conservation, 180-181
energy dissipation in flow of, 207-208
parallel, force between, 283
variable
in capacitors and resistors, 215-216
in inductors and resistors, 366-367
electric dipole
potential and field of, 73-77, 474-476
torque and force on, in external field,
477-478
electric dipole moment, 74, 473, 475
induced, 479-482
permanent, 482-483
electric displacement D, 499, 560-561
electric eels, 219
electric field
definition of, 17
in different reference frames, 243-246
of dipole, 75, 476
of Earth, 36
energy stored in, 33
of flat sheet of charge, 29
flux of, 22-26
Gauss’s law, 23-26
inside hollow conductor, 134
of line charge, 28
line integral of, 59-61
macroscopic, 488-489
in matter, spatial average of, 487
microscopic, 488
of point charge with constant velocity,
247-251
relation to 0 and p, 89
transformation of, 245, 310
units of, 17
visualization of, 18-20
electric field lines, 18, 19, 71, 72, 76-77
electric generator, 370
electric guitar, 370
electric potential, see potential, electric
electric quadrupole moment, 74, 473
electric susceptibility Xe , 490, 501, 503
electrical breakdown, 36, 100
electrical conduction, see conduction, electrical
electrical conductivity, see conductivity,
electrical
electrical conductors, see conductors, electrical
electrical insulators, 125-126
electrical potential energy, 13-16
of a system of charges, 33, 63
electrical shielding, 135
electrodynamic tether, 369
electromagnet, 320
design of, 584
electromagnetic field components,
transformation of, 310
electromagnetic force, range of, 1 1
electromagnetic induction, 343-357
electromagnetic wave, 254, 438-453
in dielectric, 507-509
in different reference frames, 452-453
energy transport by, 446-452
general properties of, 440-441
reflection of, 445, 447, 521
standing, 442-446
traveling pulse, 441
electromotive force, 209-211, 347, 357
alternating, 395
electron, 3, 5, 6, 198-204, 540-549
charge of, 8
magnetic moment of, 547
valence, 200
electron motion, wave aspect of, 199
electron orbit, 540-545
diamagnetism of, 545
magnetic moment of, 540-541
electron paramagnetic resonance (EPR),
823
electron radius, classical, 52, 545
electron spin, 546-549
angular momentum of, 546-547
electronic paper, 37
electrostatic field, 61 , see also electric field
equilibrium in, 88
electrostatic unit (esu) of charge, 8, 765
energy, see also potential energy, electrical
in alternating-current circuit, 415-418
dissipation of, in resistor, 207-208
electrical, of ionic crystal, 14-16
stored
in capacitor, 150
in electric field, 33
in inductor, 368
in magnetic field, 369
of system of charges, 11-14
energy gap, 201
equilibrium of charged particle, 88
836
Index
equipotential surfaces, 71, 131
in field of conducting disk, 140
in field of dipole, 76
in field of uniformly charged disk, 72
equivalence of inertial frames, 237, 805
equivalent circuit, 206
for voltaic cell, 211
esu (electrostatic unit), 8, 765
Faller, J. E., 11
farad (unit of capacitance), 142
Faraday, Michael, 2, 236, 314
discovery of induction by, 343-345
reconstruction of experiment by, 384
Waterloo Bridge experiment by, 380
Faraday’s law of induction, 356-357
ferrofluid, 572
ferromagnetic substances, 526
ferromagnetism, 527, 565-568
Feynman, R. R, 37, 539
field
electric, see electric field
magnetic, see magnetic field
meaning of, 245
Fisher, L. H., 348
fluctuations of current, random, 195
flux
of electric field, definition of, 22-26
magnetic, 348-351
flux tube, 349, 351
force components, Lorentz transformation of,
810-811
application of, 255-257
force(s)
between parallel currents, 283
on capacitor plate, 151, 162
conservative, 12
on electric dipole, 478
electromotive, 209-211, 347, 357, 395
with finite range, 88
on layer of charge, 30-32, 46
magnetic, 237-239
on magnetic dipole, 535-539
on moving charged particle, 255—267, 278
Foster’s theorem, 224
Frankel, R. B., 580
Franklin, Benjamin, 10, 516, 529
free and bound charge, 497-498
arbitrariness of the distinction, 506-507
free currents, 559-560
frequency modulation (FM), 455
Friedberg, R., 639
fundamental constants, 825
fuse, 219
Galili, I., 452, 464
Galvani, Luigi, 209, 236
galvanic currents, 236
galvanometer, 224, 344
Gauss, C. F., 286
gauss (unit of magnetic field strength), 282
Gaussian units, 762-768
Gauss’s law, 23-26, 80, 88
applications of, 26-30, 88, 243-245, 254,
262, 266, 488,812
and fields in a dielectric, 497^-98
Gauss’s theorem, 79-80, 100
gecko, 510
generator, electric, 370
germanium, 202
conductivity of, 195
crystal structure of, 200
resistivity of, 188
Gilbert, William, 236
Goihbarg, E., 452, 464
golden ratio, 49, 168, 231, 665
Goldhaber, A. S., 11
Good, R. H., 157, 639, 640
gradient, 63-65, 792-795
graphite
anisotropic conductivity of, 183
diamagnetism of, 546
gravitation, 3, 10, 28, 39, 163
gravitational field and Gauss’s law, 25
Gray, Stephen, 125
Griffiths, D. J., 298, 640
ground-fault circuit interrupter (GFCI), 371
gyromagnetic ratio, 541
H, magnetic field, 560-565
and B, and M inside magnetized cylinder, 565
relation to free current, 560, 561
H 2 O molecule, dipole moment of, 483
hadron, 6
Hall, E. H., 317
Hall effect, 314-317
hard disk, 571
harmonic functions, 87, 152
harmonic oscillator, 389
HC1 (hydrogen chloride) molecule, dipole
moment of, 482, 483
Heald, M. A., 298
helical coil, magnetic field of, 302
helicopters, static charge on, 102
helium atom, neutrality of, 241
helix, handedness of, 279
Henry, Joseph, 361
henry (SI unit of inductance), 361
Hertz, Heinrich, 236, 281, 314, 394
hertz (unit of frequency), 394
Hill, H. A., 11
hole, 201
Hughes, V. W., 5
hybrid car, 371
hydrogen atom
charge distribution in, 479
polarizability of, 481
hydrogen chloride molecule, dipole moment of,
482, 483
hydrogen ions, 189
hydrogen molecule, 5, 242
hydrogen negative ion, 328
hyperbolic functions, 830
hysteresis, magnetic, 569
ice, dielectric constant of, 505
ignition system coil, 372
image charge, 136-140
for a spherical shell, 159
impedance, 408-414
index of refraction, 509
inductance
mutual, 359-364
reciprocity theorem for, 362-364
self-, 364-366
circuit containing, 366-367
induction
electromagnetic, 343-357
Faraday’s law of, 356-357
inductive reactance, 396
insulators, electrical, 125-126
integral table, 826
internal resistance of electrolytic cell, 210
Index
837
interstellar magnetic field, 286, 386
invariance of charge, 241-243
distinguished from charge conservation, 242
evidence for, 241
ionic crystal, energy of, 14-16
ions, 189-198
in air, 190
in gases, 190
in water, 189-190
iron, B-H curve for, 569
Jackson, J. D., 189, 452
Jefimenko, O., 188
junction, silicon diode, 229
junkyard magnet, 321
Karlsruhe, University of, 394
King, J. G., 5
Kirchhoff’s loop rule, 207, 359
Kirchhoff’s rules, 206, 212
Laplace’s equation, 86-88, 132-134
Laplacian operator, 85-86, 799-801
lead
superconductivity of, 197
resistivity of, 197
lead-sulfuric acid cell, 209-212
Leighton, R. B., 37, 539
Lenz’s law, 351
Leyden jar, 516
Li, Y., 640
Lienard-Wiechert potential, 707
light, velocity of, definition of, 789
light-emitting diode, 220
lightning, 37
lightning rod, 153
line charge density, 28
line integral
of electric field, 59-61
of magnetic field, 287-291
linear dielectric, 490
linear physical system, 148
liquid oxygen, paramagnetism of, 525, 526, 548
lodestone (magnetite), 236, 526, 527, 565, 570
long straight wire, magnetic field of, 280
loop of a network, 207
Lorentz, H. A., 2, 236
Lorentz contraction, 261, 807
Lorentz force, 278
Lorentz invariants, 465, 811
Lorentz transformation
applications of, 247-248, 255-257
of electric and magnetic field components, 310
of force components, 810-811
of momentum and energy, 810
of space-time coordinates, 806
LR circuits, 366-367
time constant of, 367
fiQ, permeability of free space, 281
M, magnetization, 550
and B, and H inside magnetized cylinder, 565
macroscopic description of matter, 470
macroscopic electric field in matter, 488-489
maglev train, 321
magnetic bottle, 318
magnetic charge, absence of, 529
magnetic dipole
field of, 534-535
compared with electric dipole field, 535
force on, 535-539
torque on, 547
vector potential of, 531-534
magnetic dipole moment
of current loop, 534
of electron orbit, 540-541
associated with electron spin, 547
magnetic domains, 567
magnetic field, 238, 278, see also earth’s
magnetic field
of current loop, 531-535
of current ring, 299
of current sheet, 303-304
of Earth, 373
energy stored in, 368-369
of helical coil, 302
interstellar, 386
line integral of, 287-291
of long straight wire, 280
of solenoid (coil), 300-303, 338
transformation of, 310
magnetic field B, see B, magnetic field
magnetic field H, see H, magnetic field
magnetic flux, 348-350
magnetic forces, 237—239
magnetic monopole, 529
magnetic permeability /x, 563
magnetic polarizability of electron orbit, 544
magnetic pressure, 306
magnetic susceptibility Xm , 550, 563
magnetite (lodestone), 236, 526, 527, 565, 570
magnetization M, see M, magnetization
magnetogyric ratio, 541
magnetohydrodynamics, 306
magnetomechanical ratio, orbital, 541
magnetron, 419
Mania, A. J., 263
Marcus, A., 188
mass spectrometer, 317
Maxwell, James Clerk, 2, 11, 141, 236, 436
Maxwell’s equations, 436-438
Mermin, N. D., 237
metal detector, 370
methane, structure and polarizability of, 481
methanol molecule, dipole moment of, 483
method of images, see image charge
microphone
condenser, 154
dynamic, 371
microscopic description of matter, 470
microscopic electric field in matter, 488
microwave background radiation, 454
microwave oven, 419, 510
mine-shaft problem, 601
moments of charge distribution, 74, 471-474
momentum, see angular momentum
motor, electric, 319
moving charge
force on, 255-267, 278
interaction with other moving charges,
259-267
measurement of, 239-240
multipole expansion, 74, 472
muon, trajectory in magnetized iron, 582
mutual inductance, 359-364
reciprocity theorem for, 362—364
Nan-Xian, C., 643
network
alternating current, 405-414
bridge, 208, 233
direct-current, 205-207
ladder, 231
838
Index
neurons, 102
neutron, 3, 6
Newton, Isaac, 27
NH 3 (ammonia) molecule, dipole moment of,
483
nickel. Curie point of, 566
nickel sulfate, paramagnetism of, 526
Nieto, M. M., 11
niobium, 819
nitric oxide, paramagnetism of, 548
node of a network, 207
north pole, definition of, 280, 529
«-type semiconductor, 203-204
nuclear magnetic resonance (NMR), 823
nucleon, 39
nucleus, atomic, 3
octupole moment, 74, 473
O’Dell, S. L., 546
Oersted, Hans Christian, 236-237, 259, 331
oersted (unit of field H ), 775
ohm (SI unit of resistance), 186
ohmmeter, 232
Ohm’s law, 181-183, 193
breakdown of, 198
deviations from, in metals, 200
Onnes, H. K., 817
orbital magnetic moment, 540-541
oscillator, harmonic, 389
oxygen, negative molecular ion, 190
Page, L., 237
paint sprayer, electrostatic, 37
pair creation, 4
parallel currents, force between, 283
Parallel-plate capacitor, 144
filled with dielectric, 467, 489-492
parallel RLC circuit, 410
paramagnetic substances, 526
paramagnetism, 527, 540, 548
partial derivative, definition of, 64
permanent magnet, field of, 557-559
permeability, magnetic, /a, 563
permittivity, e, 497
pH value of hydrogen ion concentration, 189
phase angle in alternating-current circuit, 402,
404, 409
phosphorous, doping of silicon with, 203
photocopiers, 37
photon, 4, 460
photovoltaic effect, 220
picofarad (unit of capacitance), 142
piezoelectric effect, 511
pion, 34
Planck, Max, 2
Planck’s constant h, 546
p-n junction, 219
point charge, 21
accelerated, radiation by, 812-815
moving with constant velocity, 247-251
near conducting disk, 139
starting or stopping, 251-255
Poisson’s equation, 86 , 89
polar molecules, dipole moments of, 483
polarizability
magnetic, of electron orbit, 544
of various atoms, 481-482
polarization, frequency dependence of, 504
polarization density P, 484, 498, 501, 503
polarized matter, 483-489
polarized sphere, electric field of, 492-495
pollination, by bees, 509
positron, 3
potential
coefficients of, 148
electric, </>, 61-73, 86-89
of charged disk, 69
of charged wire, 67
derivation of field from, 65
of electric dipole, 73-74, 475
of two point charges, 66
vector, 293-296
of current loop, 531-534
potential energy, electrical, 13-16
of a system of charges, 33, 63
power
in alternating-current circuit, 415-418
dissipated in resistor, 208
radiated by accelerated charge, 814
power adapter, 420
power-factor correction, 420
Poynting vector, 448-452
precession of magnetic top, 821, 822
Press, F., 380
Priestly, Joseph, 10
proton, 3
decay of, 6
and electron charge equality, 5
magnetic moment of, 822
/ 7 -type semiconductor, 203—204
Q, of resonant circuit, 392, 402
quadrupole
moment, 74, 473
tensor, 514
quantization of charge, 5-7, 242
quantum electrodynamics, 2
quark, 6, 35
quartz clock, 511
radiation by accelerated charge, 812-815
radio frequency identification (RFID) tags, 454
railgun, 319
random fluctuations of current, 195
range of electromagnetic force, 1 1
rare-earth magnets, 573
rationalized units, 767
RC circuit, 215-216
time constant of, 216
reactance, inductive, 396
reciprocity theorem for mutual inductance,
362-364
recombination of ions, 190
refractive index, 509
regenerative braking, 371
relaxation method, 153, 174
relaxation of field in conductor, 217
relaxation time, 217
relay, electric, 320
remanence, magnetic, 569
resistance, electrical, 183-187
resistances in parallel and in series, 206
resistivity, 186
of various materials, 188, 196
resistor, 205, 207
resonance, 418
resonant circuit, 388-394
damping of, 391-394
critical, 394
energy transfer in, 392
resonant frequency, 400
retarded potential, 329
Roberts, D., 211
Index
839
Rodrigues, W. A. Jr., 263
Romer, R. H., 359
Rowland, Henry, 259, 314, 315, 317
Rowland’s experiment, 315
RLC circuit
parallel, 410
series, 389, 398
Sands, M., 37, 539
saturation magnetization, 565
scalar product of two vectors, 12
Scott, G. K., 305
seawater, resistivity of, 188, 190, 196
second (as Gaussian unit of resistivity), 187
self-energy of elementary particles, 35
self-inductance, 364-366
circuit containing, 366-367
semiconductors, 126, 195, 200-204
«-type, 203-204
p-type, 203, 204
Semon, M. D., 296
series RLC circuit, 389, 398
shake flashlight, 370
sheets of charge, moving, electric field of,
243-245
shielding, electrical, 135
SI units, 762-768
derived, 769
Siever, R., 380
silicon, 195, 200-204
band gap in, 201
crystal structure of, 200
slope detection, 455
smoke detector, 219
Smyth, C. P., 505
sodium and chlorine ions in water, 190
sodium chloride crystal
diamagnetism of, 526
electrical potential energy of, 14-16
free and bound charge in, 507
sodium metal, conductivity of, 198-199
solar cells, 220
solenoid (coil), magnetic field of, 300-303, 338
speakers, 321
spherical coordinates, 792
spin of electron, 546-549
sprites, 219
St. Elmo’s fire, 37
standing wave, electromagnetic, 442-446
Starfish Prime, 318
statvolt (Gaussian unit of electric potential), 61
Stokes’ theorem, 92-93, 100
storage battery, lead-sulfuric acid, 209-212
supercapacitor, 154
superconductivity, 197, 817-820
superposition, principle of, 10
applications of, 25, 147, 207, 245, 301, 442,
490,492
surface charge
on current-carrying wire, 188-189, 263,
452
density, 129
distribution, 29
surface current density, 303
surface integral, definition of, 23
surfaces, equipotential, see equipotential
surfaces
surfactant, 510
susceptibility
electric Xe , 490, 501, 503
magnetic Xm, 550, 563
symmetry argument, 21
synchrotron radiation, 815
Taylor, J. R., 296
Taylor series, 827-828
television set, 318
temperature, effect of
on alignment of electron spins, 548-549
on alignment of polar molecules, 503
on conductivity, 195-197
Tesla, Nikola, 286, 419
tesla (SI unit of magnetic field strength), 280
Thevenin’s theorem, 213-215, 225
three-phase power, 419
torque
on current loop, 332, 547
on electric dipole, 477, 478
transatlantic telegraph, 227
transatlantic telegraph cable, 217
transformation, see Lorentz transformation
transformer, 372
transistor, 220
triboelectric effect, 36
trigonometric identities, 829
uniqueness theorem, 132-133
units, SI and Gaussian, 762-768
conversions, 774-777, 789-790
formulas, 778-788
vacuum capacitor, 467
valence band, 201-204
valence electrons, 200
Van Allen belts, 318
Van de Graaff generator, 182, 209, 211
van der Waals force, 510
Varney, R. N., 348
vector identities, 827
vector potential, 293-296
of current loop, 531-534
vector product (cross product) of two vectors,
238
volt (SI unit of electric potential), 61
Volta, Alessandro, 209, 236
Voltaic cell, 209
equivalent circuit for, 211
voltmeter, 224
Waage, H. M., 530
Walker, J., 35
War of Currents, 419
water
dielectric constant of, 505
ions in, 189-190
pure, resistivity of, 188, 196
water molecule, dipole moment of, 483
watt (SI unit of power), 208
wave, electromagnetic, see electromagnetic wave
weber (SI unit of magnetic flux), 357
Whittaker, E., 500
Williams, E. R., 11
wire
charged, potential of, 67
magnetic field of, 280
work, by magnetic force, 572
Zia, R. K. P., 546
Derived units
kgm
newton (N) = —r—
s-
joule (J) = newton-meter
coulomb C
ampere (A) = -= —
second s
volt (V)
farad (F)
ohm (£2)
watt (W)
tesla (T)
henry (H)
joule kg m 2
coulomb C s 2
coulomb C 2 s 2
volt kg m 2
volt kg m 2
ampere C 2 s
joule kg m 2
second s 3
newton kg
coulomb • meter/second C s
volt kg m 2
ampere/second C 2
Maxwell’s equations
3B
curl E-
3 1
3E
curlB = [a 0 e 0 —
dt
P
div E = —
fo
div B = 0
MoJ
Fundamental constants
speed of light
c
2.998 ■ 10 8 m/s
elementary charge
e
1.602-10“ 19 C
4.803- 10“ 10 esu
electron mass
m e
9.109 • 10 -31 kg
proton mass
m p
1.673 • 10- 27 kg
Avogadro’s number
Na
6 .022- 10“ 23 mole -1
Boltzmann constant
k
1.381 • 10 -23 J/K
Planck constant
h
6.626 ■ 10 -34 J s
gravitational constant
G
6.674-10“ 11 m 3 /(kg s 2 )
electron magnetic moment
de
9.285-10 -24 J/T
proton magnetic moment
dp
1.411 • 10 -26 J/T
permittivity of free space
eo
8.854-10“ 12 C 2 s 2 /(kg m 3 )
permeability of free space
dO
1.257 -10 -6 kg m/C 2
Divergence theorem
/'■*- /■
F • da = I div F dv
surface volume
Stokes’ theorem
f Ads = /■
(A • ds = J curl A • da
curve surface
Gradient theorem
02 — 01
grad 0 • ds
Vector operators
Cartesian coordinates
ds = dx x + dy y + dz z
,3 ,3 „ 3
V = x-h y-h z—
dx dy dz
3/ „ 3/ „ 3 / „
V/= — x+ — y+ — z
3x dy dz
d A x 3A V d A z
V ■ A = —+ ^ + —
3x 3y 3z
V x A =
/3Aj _ 3Ay\ /3A* _ 3A,\ /3Ay _ 3AA
\ dy dz J X+ { dz dx J y+ { dx dy J
3 2 f , 3 2 f
3 2 f
V f~y x 2 + ^2 + ^2
Cylindrical coordinates
ds = dr r + rd8 § + dzi
„ .3 -1 3 .3
3r r 3d dz
df „ 1 3/ - 3/ „
J 3r r 30 3z
1 3(rA r ) 1 3A 0 3 A z
V ■ A = - +-- + —
r dr r dd dz
V x A =
/ 1 3A Z 3A fl \ /
(r 3fl 3z / + \
, 1 3 / 3/\ 1
V/ “ ~rTr YTr) + 7l
d A r
dz
dA z
dr
)§+
/ r V dr as J
3 2 f
d 2 f
2 3d 2 dz 2
Spherical coordinates
rfs = rfr r + r d6 6 + r sin 0 dip <j>
,3 -13 - 1 3
V = r + d-+0-
3r r a9 rsind dtp
df „ 13 f ~ 1 df -
V/ = — r + - — d +-- <1
dr r d8 rsind dtp
v A= l^) +
r 2 3r rsind
1 3(Ag sin#) | 1 3A 0
rsind dtp
ae
+
VxA- 1/ 1 3A r 3(rA 0 )
3d 30 / r \sind dip dr
a 2 f
rsin0 \
) §+ ~r
1 / d(rAg) _ aAr
\ dr 3d
v 2 / =
>if^ + _L>L4) +
r 2 dr \ dr J r 2 sind 3d \ dd J
1
r 2 sin 2 d dip 2
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Books to Borrow
Open Library
TV News
Understanding 9/11